Judging Trends in Income Inequality in the U.S.: Do the Measures Matter?

ABSTRACT This paper uses Piketty, Saez, and Zucman's income measures that are constructed to be consistent with distributional national accounts from 1966 to 2019. Three inequality indexes and two strategies for estimating them were evaluated using the pre‐tax and post‐tax income measures. The indexes were: the Gini coefficient, mean logarithmic deviation, and the Theil index. The results imply that the conclusion on trends in income inequality depends on the measure used. Using the index recommended as reflecting changes in the full distribution of income, the mean logarithmic deviation, both model‐based estimates and distribution‐free estimates suggest income inequality at the end of the PSZ sample (2019) was about the same as it was in 1966.