It has become commonplace to use top 1 percent shares of market income as a shorthand measure of inequality, and as an argument for greater taxes on higher incomes and/or larger transfer payments to the bottom 90 percent. This paper finds the data inappropriate for such purposes for several reasons:
- Excluding rapidly increased transfer payments and employer-financed benefits from total income results in exaggerating the rise in the top 1 percent’s share between 1979 and 2010 by 23 percent because a growing share of other income is missing.
- Using estimates of the top 1 percent’s share of pretax, pretransfer income (Piketty and Saez 2003) as an argument for higher tax rates on top incomes or larger transfer payments to others is illogical and contradictory because the data exclude taxes and transfers.
- Using highly cyclical top 1 percent shares as a measure of overall inequality leads, paradoxically, to describing most recessions as a welcome reduction in inequality, because poverty and unemployment rates typically rise when the top 1 percent’s share falls, and fall when the top 1 percent’s share rises.
- Top 1 percent incomes are shown to be extremely sensitive (“elastic”) to changes in the highest tax rates on ordinary income, capital gains and dividends. Although estimates of the elasticity of ordinary income for the top 1 percent range from 0.62 (Saez 2004) to 1.99 (Moffitt and Wilhelm), those estimates fail to account for demonstrably dramatic responses to changes in the highest tax rate on capital gains and dividends.
I estimate that more than half of the increase in the top 1 percent’s share of pretax, pretransfer income since 1983, and all of the increase since 2000, is attributable to behavioral reactions to lower marginal tax rates on salaries, unincorporated businesses, dividends and capital gains. After reviewing numerous data sources, I find no compelling evidence of any large and sustained increase in the inequality of disposable income over the past two decades.
