Elizabeth Warren and Bernie Sanders are making income inequality a focus of their presidential campaigns, a priority that had led them to demonize wealthy Americans and to propose ever-more imaginative tax increases on those in the top income brackets. Sanders talks about the level of inequality as “grotesque and immoral,” while Warren claims the U.S. economy only works for a “thinner and thinner slice at the top.” But with workers’ wages rising under President Trump at rates not seen in a decade, is income inequality actually increasing?

A September report by the U.S. Census Bureau based on The American Community Survey (ACS) appeared to boost Sanders’ and Warren’s claims by finding that income inequality “was significantly higher” in 2018 than in 2017.

Another Census Bureau report issued two weeks earlier reached a far different conclusion, however. By two separate measures, The Current Population Survey’s Annual Social and Economic Supplement (CPS) found that inequality decreased in 2018. By one measure, it found a decrease that was “not statistically significant.” By another, more complete, measure, it found a “significant” decrease.

There are some timing differences between the surveys, and the CPS has a wider margin of error than the ACS. With conflicting results, it can be difficult to discern which survey is more reliable. Fortunately, the Census Bureau provides guidance, recommending the CPS survey “as the data source for national estimates” because it produces “more complete and thorough estimates of income and poverty.”  

Despite this recommendation, numerous media outlets -- from The New York Times and Politico to USA Today and The Washington Post -- ran articles ignoring the CPS and citing the ACS’s finding that income inequality increased in 2018,  even claiming that it increased to its highest level in at least 50 years. This was odd because the ACS goes back only 12 years to 2006. It’s the CPS that goes back 50 years (with the highest level of inequality in 2017, not 2018.)

One would think the CPS report was at least worth mentioning. Even the ACS press release identified the CPS as the “leading source for national data” on income and poverty. In any event, both reports calculated income inequality using median household income and applying the Gini Index, a widely accepted statistical measure of income inequality. The Gini Index measures inequality on a scale from 0 to 1, so small changes can make a big difference. The CPS found a .003 decrease in inequality (from .485 to .482) while the ACS found a .003 increase (from .486 to .489).

Of course, the result is only as good as the data. According to the Census Bureau, the CPS “income questions are much more detailed” than those in the ACS and “provide more comprehensive coverage of all potential income sources.” This helps explain why the CPS found higher median household income ($63,179) than the ACS ($61,937) and a slight decrease in income inequality.

A drawback of this approach is that it treats all households the same even though households come in different sizes. For example, $30,000 in income impacts a single-person household far differently than a family of four.  The CPS report addresses this defect by additionally using an “equivalency” measure of inequality that considers “the number of people living in the household and how these people share resources and take advantage of economies of scale.” Under this measure, a household of four would need $65,000 in income to be considered equivalent to a single-person household earning $30,000.

Applying the Gini Index under the equivalency measure, the CPS found a “statistically significant” decline in income inequality for 2018 of .007 (from .471 to .464). In fact, it shows that the share of household income increased 3.4% for the lowest quintile (20%) of earners, 2.4% for the next lowest, and 1.5% for the middle quintile. The top 20% of earners was the only quintile to experience a decrease.

So, employing the equivalency measure, the data source “the Census Bureau recommends” found that income inequality is moving in exactly the opposite direction that Warren and Sanders claim. In fact, last year it decreased significantly.

That inequality declined in 2018 should come as no surprise. Thanks in great part to tax cuts and regulatory reductions, by December of 2018 the unemployment rate had declined to 3.9% – the lowest year-end rate in almost 20 years. Labor participation increased from 62.7% in December 2017 to 63.1% in December 2018 as more people came off the sidelines to find jobs. It was the first year-end over year-end increase since 2006. In March, the number of job openings exceeded the number of people unemployed for the first time since the government began reporting the data.  By year end, there were 1.2 million more job openings than people unemployed.

As a result, for each of the last five months of 2018, hourly wages increased more than 3% on a yearly basis -- for the first time in nearly a decade. In December, wages increased 3.3%, the best year-end increase since 2008. Workers’ wages drove this increase, rising 3.5% (also the best since 2008) while supervisors’ wages increased only 2.5%. Traditionally low-wage retail workers ended the year with a 5.1% increase, their best year-end since 1981.

Income inequality decreased in 2018 as the economy created jobs, increased the demand for employees and drove meaningful wage gains for working-class Americans. Ignoring that decrease won’t make it go away.