
Piketty and Saez (2003) found a pronounced U-curve pattern of American income inequality since 1917, displaying a precipitous decline during World War II to a level that would hold until 1980. We offer revisions to their income inequality estimates prior to 1960 with three important findings. First, Piketty and Saez overstate inequality levels in this period. Second, the decline during WWII was smaller than depicted. Third, the Great Depression, rather than WWII, played the more significant role. These findings indicate a need to reevaluate commonly held assumptions about the evolution of inequality during the period of the ‘Great Leveling,’ as well as the nature of its posited relationship to tax policy.
This is from Vincent J. Geloso, Phillip Magness, John Moore, and Philip Schlosser, “How Pronounced in the U-curve? Revisiting income inequality in the United States, 1917-1960,” The Economic Journal, March 8, 2022.
The Economic Journal is a prestigious British journal that also charges you a lot to buy the paper. So instead, you can read an earlier version that was published on SSRN. It’s here.
Here’s an excerpt from that earlier version:
The purpose of this paper, we must emphasize, is not to assert that the United States was an exception to century-long international patterns in inequality. Neither do we question the idea that top income shares fell then rose over the 20th century. We accept the general occurrence of the “great levelling,” subject to further investigation of its shape and magnitude. We show instead that the left-side of the PS [Piketty/Saez] U-curve likely overstates the original height and ensuing decline of top income shares, with implications for distributional pattern[s] over time and for other derivative works that rely upon the PS series. Our adjustments do not change the fact that inequality fell between 1929 and 1945. What they do is attenuate this decline so that, when combined with other proposed corrections to the right-side of the U-curve, we see a much shallower “tea saucer” of inequality in the United States over the the course of the twentieth century. We confirm an inequality peak in 1928-1929, which is actually much more acute to those years than in PS. However, the “great levelling” (an expression we borrow from Lindert and Williamson [2016]) becomes a comparatively gradual story where the Great Depression is a more readily visible factor. This draws into question numerous inferential claims that wartime income tax policy and its subsequent entrenchment as a permanent feature of the mid-century tax system played a major role in the observed levelling.
Past editors of The Economic Journal included Francis Y. Edgeworth (from 1891 to 1911) and John Maynard Keynes (from 1912 to 1944.)
HT2 Phil Magness.
READER COMMENTS
Alan Reynolds
Mar 12 2022 at 6:47am
The authors kindly mention my 2006 book and a 2012 paper on this topic, yet my analysis said little about the long-term history of the Piketty-Saez data series except to note that their prewar and postwar figures were not strictly comparable. My analysis largely focused to the high sensitivity of their estimates to several momentous changes after 1986 in marginal tax rates on corporate and individual tax rates on income and capital gains reported (or not) on individual or corporate tax returns. I also argued that the Piketty-Saez concept of pretax, pretransfer income was a meaningless measure of relative living standards.
“I raise three principal objections to using data from the top 1 percent of tax returns as a measure of changes in income inequality among all households. I argue, first, that the Piketty and Saez estimate of total market income −the denominator of the ratio of top percentile income to total income− understates actual incomes of the “other 99 percent” by huge and growing amounts. Second, the top 1 percent’s share is too cyclical to be a meaningful measure of inequality, since reductions in the top 1 percent’s share are typically associated with higher rates of poverty and unemployment while increases in the top 1 percent’s share are associated with falling poverty and unemployment. To adopt top percentile shares as a measure of inequality requires accepting the paradoxical conclusion that inequality “improves” when poverty worsens, and vice-versa. Third, and perhaps most important, the top 1 percent’s share is shown to be extremely sensitive to changes in marginal tax rates on salaries, business income, dividends, and capital gains. As a result, behavioral responses to higher or lower tax rates have been frequently misinterpreted as actual changes in top incomes rather than as changes in the way incomes are or are not reported on individual or corporate tax returns.” WorkingPaper-9.pdf (cato.org)
Cyril Morong
Mar 12 2022 at 7:13pm
Does this paper use household income or individual income?
Mark Brady
Mar 13 2022 at 3:34pm
David writes, “The Economic Journal is a prestigious British journal that also charges you a lot to buy the paper.” Indeed, and I’m pretty sure that it is also the first scholarly journal specifically devoted to economics to be published in Britain (1891). Readers may remember that the following year it published Carl Menger’s “On the Origin of Money” in Vol. 2, No. 6 (June 1892). Unfortunately it printed the author’s name as “Karl Menger,” a source of confusion ever since. 🙁
Mark Brady
Mar 13 2022 at 3:40pm
I think this is a more recent version of their paper.
https://academic.oup.com/ej/advance-article/doi/10.1093/ej/ueac020/6544663?searchresult=1
Comments are closed.