January 31, 2024
Just as there seems to be a limitless demand for scholarly or pseudo-scholarly pieces that deny global warming, there also is an insatiable demand for pieces that deny the rise in income inequality over the last four decades. The latest entry in this area was an article by Gerald Auten and David Splinter (AS) arguing that income tax data do not show the rise in inequality often claimed.
The main target for Auten and Splinter’s analysis is a series of papers by Thomas Piketty, Emanual Saez, and Gabriel Zucman (PSZ). These papers, also based on income tax data, show a sharp rise in income inequality from 1979 to the present, with the share of the richest one percent going from 13 percent to 19 percent, a rise of six percentage points. (It is worth noting that these numbers exclude income from capital gains, which does not count as income in the GDP accounts.) By contrast, Auten and Splinter find an increase in the share of the top one percent of just one percentage point over this period.
In recent weeks, the Auten and Splinter piece has been highlighted in an article by Dylan Matthews in Vox and a column by Eduardo Porter in the Washington Post. The takeaway from both pieces is that our concerns about inequality have been grossly exaggerated.
Inequality: Income Tax Data and Beyond
There are many differences in the way that AS and PSZ analyze the tax data, but the largest single difference is how they allocate unreported income. (A paper from Brookings, by William Gale, John Sabelhous, and John Thorpe, provides an excellent analysis of the differences between the AS and PSZ studies.) PSZ assume that unreported income is distributed pretty much the same way as reported income. AS relies on I.R.S. audit studies and concludes that a grossly disproportionate share of unreported income goes to the bottom 99 percent of tax filers.
In response, PSZ makes the obvious point that well-hidden income will likely not show up in audit studies. (Remember, these people are committing a crime by not paying taxes they owe. They might need to keep their income hidden to stay out of jail.) Furthermore, the rich will be best positioned to pay for complex and high-cost tax evasion strategies.
But even if there may be some basis for ambiguity in the income tax data, we have other data sources that tell pretty much the same story as PSZ. Most obviously we have the Current Population Survey (CPS) that the Censis Bureau fields every month. This survey top codes income at roughly the cutoff for the richest one percent of the population, so it can’t tell us directly how much income they are receiving. However, it does tell us what 99 percent of the population is receiving.
The CPS also allows us to see the wide gap between productivity and the pay of the median worker that has opened up since 1979. The Labor Department’s monthly survey of employers, the Current Employment Statistics, also shows the same gap.
The Social Security Administration provides an even more comprehensive source of data with its series on annual wages. The great benefit of this data series is that it is not top-coded. An analysis of these data from Elise Gould and Jori Kandra shows an increase in the top 1.0 percent’s share of wage income from 7.3 percent in 1979 to 12.9 percent in 2021, a rise of 5.6 percentage points. This is very close to the 6.0 percentage point rise in the PSZ analysis.
This sort of increase in high-end pay is also consistent with the explosion in the ratio of CEO pay to the pay of ordinary workers. This ratio went from just over 20 to 1 in the 1960s to more than 340 to 1 in 2022. It is worth remembering the exorbitant pay doesn’t go just to the CEO. If the CEO is getting $30 million then the CFO and other top officers are likely getting $10 to $15 million, and even third tier execs can get $1-2 million. The story looks very different if we were back in a world where the ratios were 20 to one, and the CEO might be getting $2 to $3 million a year.
It’s true these series only measure inequality in wage income, but labor compensation is almost 68 percent of total national income. That leaves capital income with just 32.0 percent of the total. You would need some pretty crazy movements towards greater equality in non-wage income to qualitatively change the picture shown in the wage data.
In fact, there has been a modest shift from labor income to capital income with the labor share falling from 71.8 percent in 1979 to 68.0 percent in 2022, the most recent year for which full data are available. Since capital income is more unequally distributed than labor income, this shift would increase the amount of inequality found in the wage data. (Many people wrongly believe that the story of inequality is a shift from wages to profits. In fact, most of the story of inequality over the last four and a half decades has been within the wage distribution.)
In short, the sort of rise in inequality found by PSZ is consistent with the rise in inequality shown in a number of other data sets. There will always be some room for judgment calls in assessing any specific data set, but when they all seem to be going in the same direction, it is difficult to contest the conclusion. We have seen a huge increase in inequality since 1979.
Just as there will always be a market for studies that purport to show that global warming is not happening, there will likely always be a market for analyses that deny the growth in inequality we have seen over this period. While we should always be prepared to question accepted wisdom, it is important to recognize how firm the basis for that accepted wisdom is. In the case of the rise in inequality, it is rock solid.