Earlier this week I noted that we were on the verge of a big debate over how to tax the rich, thanks to a new book by Emmanuel Saez and Gabriel Zucman and a column in the New York Times promoting it. The Times piece featured this shocking chart depicting an almost flat tax code when federal, state, and local taxes are combined:
David Splinter, an economist who’s criticized Saez and Zucman’s income data in the past, has now posted a short but detailed response. (Splinter works with the Joint Committee on Taxation but is speaking for himself here.) Basically, it’s tricky to nail down how much income people actually earn, and thus to use the right denominator when calculating tax rates — and Saez and Zucman’s decisions tend to make the tax code look much less progressive than it really is by increasing pre-tax income for the rich and reducing it for the poor.
Both Splinter and Saez/Zucman (and their past coauthors) make an attempt to include income that wasn’t reported to the IRS, which is important because “wage and business income in national income data exceed amounts reported on tax returns by over a trillion dollars” — but Splinter assigns less of that mysterious income to the rich. Splinter relies on “representative IRS audit studies,” while Saez-Zucman make an “ad hoc assumption that such income is proportional to reported source-specific income.”
A similar phenomenon happens with untaxed retirement income: Splinter doesn’t count “rollovers,” which he says merely represent “assets shifting between accounts,” as income. Splinter further doesn’t “deduct most payroll taxes” before calculating pre-tax income, which would reduce the income of the poor and middle class disproportionately.
Splinter further notes that while Saez and Zucman cite their own past work with Thomas Piketty (“PSZ”) as the source for the tax-rate estimates for 2018 that were ultimately presented in the New York Times, “the 2018 tax data files used for PSZ estimates will not be available until about a year after their conference draft was posted. Hence, these estimates necessarily use a different method than the cited source.” Splinter “forecasts” 2018 tax-rate estimates himself based on the PSZ paper and finds that they do not quite match the numbers that ended up in the Times, especially at the very top. He also applies the same method to his own past work with Gerald Auten to show how much more progressive the tax code looks under his method.
Here’s a nice chart showing how it all comes together, though unfortunately it doesn’t include the tax rates for the 400 wealthiest taxpayers. Splinter’s method shows a much more progressive tax code, both in 2014 and in the 2018 forecasts:
By the way, Splinter does not adjust the data to account for refundable tax credits, an aspect of the tax system that disproportionately helps the poor and working class and would make the tax code look even more progressive by reducing tax rates for the lower half of the income distribution by “up to 3 percentage points.”
Also worth tossing in here: Jason Furman — yes, from the Obama Council of Economic Advisers — put together some numbers from other sources, and also found a far more progressive tax code than the NYT presented:
The standard data shows that the tax system is overall progressive. This chart combines CBO estimates for federal taxes with ITEP estimates for state & local taxes. Federal income taxes highly progressive, when you add in payroll/state/local/etc. is still progressive but less so. pic.twitter.com/WTOgm58Fyo
— Jason Furman (@jasonfurman) October 7, 2019