How Pronounced Is the U-Curve? Revisiting Income Inequality in the United States, 1917-1945

The Great Overestimation: Tax Data and Inequality Measurements in the United States, 1913–1943

“…However, the data presented here can also shed some light on these methodological assumptions and the problems they may pose. Geloso et al () emphasize that the bulk of the methodological problems with the necessary adjustments to the IRS data were related to the adjustment for missing filers, the adjustment to convert net income into AGI, and the denominator of total income used by Piketty and Saez (, ). We are unable to speak to the point of the missing filers as we do not have access to the data transformations that Frank () made to adjust for missing filers, which prevents us from assessing the importance of the adjustment of IRS data for the top 10% income share (the adjustment does not affect higher‐income fractiles).…”

“…However, our data can shed further light on the problem of the adjustment from net income to AGI. For the period from 1926 to 1938, Geloso et al () argue that the adjustment Piketty and Saez made to deductions overstates the level of top 1% income concentration by 6% (1 percentage point) at the national level. For Delaware, during the period of available data, the same adjustment made by Piketty and Saez understates the level of inequality by 3% on average (<1 percentage point).…”

“…This is a logical impossibility. In their work, Geloso et al () point out that the Piketty–Saez income denominator was implausibly low and that mild accounting correction to the denominator brought down the top income shares estimated by Piketty and Saez (). The state of the national‐accounts data does not permit us to replicate the efforts of Geloso et al () for Delaware (or Wisconsin), but the evidence from Delaware's state income tax reinforces the case that interwar inequality estimates are overstating the level of income inequality.…”

“…In their work, Geloso et al () point out that the Piketty–Saez income denominator was implausibly low and that mild accounting correction to the denominator brought down the top income shares estimated by Piketty and Saez (). The state of the national‐accounts data does not permit us to replicate the efforts of Geloso et al () for Delaware (or Wisconsin), but the evidence from Delaware's state income tax reinforces the case that interwar inequality estimates are overstating the level of income inequality. If we import the national‐level corrections that Geloso et al () made to the income denominator for Delaware and Wisconsin (i.e., augmenting the personal‐income figure by a fixed factor), the income share for the richest 1% falls by roughly 3.9% points for Delaware and 1.0 for Wisconsin.…”

“…The accuracy of the discretionary decisions in the steps of the Piketty–Saez calculations can be debated (Geloso et al ). However, since we are merely interested in isolating the effects of the choice of data source in the main narrative of inequality, we take these steps as a given for the current study.…”