clock menu more-arrow no yes mobile

Filed under:

Inequality really is rising, no matter how you fuss with the data

Last week, economist Pavlina Tcherneva made a chart, based on data from Emmanuel Saez and Thomas Piketty, showing that rich Americans have started capturing more and more of the gains from economic booms. The chart went viral. We at Vox even made a video about it (above). And then something enormously predictable happened. Any time a depiction of growing inequality in the United States becomes popular, people who believe that increased inequality isn't bad, even if it's real, start arguing that it isn't real either. Scott Winship is one of the brightest lights of that school of thought, so he presented a long series of quibbles with both the Saez/Piketty data and Tcherneva's presentation of it which he framed as a debunking of their claims.

But look past the framing and you'll see that even Winship himself agrees that "income inequality is at staggering levels in the US, and that income concentration at the top has probably risen." After all the quibbling is said and done, Winship and Tcherneva don't really disagree — it's more that Tcherneva thinks the inequality is important and wants to draw attention to it, while Winship does not and thus opposes Tcherneva-style framings that dramatize growing inequality. At the end of the day, whether inequality is important is not something charts or data can prove. But we can help you understand what the data disputes are about.

Tax units versus households

One issue Winship raises is that the Saez/Piketty data is based on "tax units," an IRS term of art that does not correspond to a more natural social phenomenon like households. They do this because examining tax return data (as opposed to survey data from the Census Bureau or elsewhere) is the only way to get a handle on the incomes of the very rich. And the only way to work with tax return data is to look at tax units. Winship correctly points out that the average tax unit is considerably poorer than the average household — a teenager living at home and working a part-time job is going to be a very poor tax unit.

But though this is true, it has nothing to do with the point the chart is illustrating which is change over time. It used to be that most of the income gains during a recovery accrued to the bottom 90 percent of tax units. More recently those gains have gone to the top 10 percent of tax units. It is true, but irrelevant, that tax units are not households.

Taxed versus untaxed capital gains

A related point about the use of tax data is that it somewhat overstates the unequal distribution of capital gains. That's because capital gains income derived from the sale of a house or for some certain retirement accounts is normally not taxed. Middle class individuals hold a very large share of their savings in these kind of tax-preferred vehicles, and the income garnered from them does not show up in IRS data. Consequently, both middle class and wealthy families have higher capital gains incomes than is shown in IRS data. But in percentage terms, the IRS data understates middle class income by more than it understates incomes of the wealthy.

This, too, is true but not relevant to the point at hand. The chart illustrates a point about the change over time in how economic booms are distributed, not a point about the absolute level of various classes' incomes. Tax data overstates the level of inequality in capital gains income at any given time, but that doesn't change the reality that inequality has grown.

Recoveries versus business cycles

(Thomas Piketty and Emmanuel Saez)

Winship's most illuminating point is his complaint that by looking only at periods of economic growth rather than entire business cycles including recessions, Tcherneva has made an excessively dramatic looking chart. The issue is that stock market losses during recessions are very big, and inequality diminishes during this period. If you look at the full Saez/Piketty data series you can see this clearly — during stock market crashes, the top 10 percent's share of income falls.

That said, these are simply two different ways of making charts out of the exact same data. Tcherneva's is a more dramatic presentation, and the version of Tcherneva's chart that Joe Posner made for the video is even more dramatic, since he's a very talented visual designer. The dispute here is not actually a dispute about what the tax data says, it's a dispute about whether finding a dramatic new illustration of the tax data is a valuable undertaking or a dangerous one.

Taxes and transfers

One issue often lost in popular discussions of income and poverty trends is the difference between data that accounts for taxes and transfers and data that excludes it. The IRS data is about what people say on their tax returns — i.e., it is income before subtracting taxes and before adding many kinds of government payments. Since rich people pay a lot of taxes and many low income people are retirees who get Social Security checks, income is more equally distributed after taxes and transfers than before.

Both kinds of data are important. If you want to know about living standards for people in the bottom third of the income distribution, it is crucially important to try to account for the value of government programs. The Affordable Care Act, for example, is going to give deeply discounted health insurance policies to many working class Americans. That is a very real and genuinely important increase in their standard of living. It is not, however, an increase in their incomes as measured by the IRS. That said, reasonable people will also be interested in facts about pre-tax income. If you are interested in questions of macroeconomic management, corporate governance, or labor market policy then you are going to want to know about the distribution of pre-tax income.

An unviable alternative

(Scott Winship)

Winship suggests that the above chart would undermine the story of growing inequality, while also offering a more accurate picture. This chart excludes capital gains income entirely, and also only examines tax units with taxable income — i.e. the jobless don't count. And it is true that if you ignore the economy's tilt in favor of owners and against workers, and also ignore 15 years of persistent labor market weakness, then the overall economic picture looks a lot brighter. Other than that, Mrs. Lincoln, how was the play?

A robust conclusion

The bottom line is that while there is lots of room for arguing about the details of data selection, the conclusion that inequality is growing is not an eccentric discovery made by two French economists working with a particular source of income data. The broad trend is clear from a diverse set of data. Median household income growth has badly lagged per capita GDP growth, corporate profits as a share of national income have risen, and stock markets have reached record highs.

Outside the sphere of political debate, you also see the real world impact of inequality. Merrill Lynch recommends an investment strategy to its clients based on the growing economic clout of plutocrats, Singapore Airlines is now selling $18,400 first class cabin tickets, and observers think Apple is going to start selling a $10,000 watch. Conversely, Walmart is now primarily worried about competition from dollar stores. The executives at these companies are not hysterical liberals trying to drum up paranoia about inequality, they are trying to respond to real economic conditions — conditions that have entailed very poor wage growth paired with decent returns for those proserous enough to own lots of shares of stock.

Looping back to the beginning, the most striking fact about the entire dispute is that Winship himself does not disagree that inequality is at a very high level and that it has risen since the Reagan Revolution. He simply thinks it's wrong to obsess over this, and therefore that it's wrong to try to think up ways to dramatize it. But if you think that America is slipping into a doom loop of oligarchy then naturally you will want to dramatize the trend. Tcherneva's chart does that very well, which is exactly why it's so controversial.