Markus Schneider and I have a brand spanking new working paper just appeared on the Levy Economics Institute webpage.
The paper addresses the following puzzle. Most economists are familiar with the U-shaped pattern followed by the share of top income earners as documented by Piketty and Saez (PS, 2003, 2006). Top incomes captured about 20% of US GDP just before the Great Depression. Then, the share of top incomes declines to 7% or 8% in the 1960s and ’70s, to steadily climb starting in the 1980s, up to 20% again just before the Great Recession.* Hence, the observation that inequality in the US has decreased from the 1930s to the 1970s, and then increased afterwards. If one looks at the Gini index, however, no U-shaped pattern can be found: the US Gini has been steadily rising since the 1940s, suggesting growing inequality ever since.
In an interesting Proceedings of the AMS paper, Jantzen and Volkert (JV, 2012) have proposed a decomposition of the Gini index into two indices representing, respectively, inequality at the top of the income distribution (similarly to the top income share in PS), and inequality at the bottom – that is, roughly, the distributional gain of pulling oneself out of poverty. Markus and I use IRS Adjusted Gross Income data from 1921 to 2012 to estimate JV’s indices for the period. We find that the increase in the Gini can be decomposed into two periods: from the 1940s to the late 1970s, the Gini increases because inequality at the bottom statistically increases more than inequality at the top. Meaning that distributional gains accrue to “working class” and “middle class” incomes relative to top incomes, but mostly at the expenses of the very poor.
From 1981 to 2012, instead, the increase in the Gini is driven by increasing inequality at the top: top incomes gain at the expenses of everybody else. We document that the relative index mirrors closely the pattern followed by the share of top incomes in PS. Finally, we argue that the two periods can be ranked in relative welfare terms, once economic growth is factored in. The more recent period appears to be welfare reducing for over 93% of income earners, while the welfare effects of increasing inequality at the bottom in the earlier period are ambiguous.
Read the paper here.
*PS have a series of papers arguing that the reason has to be found in the evolution of the top marginal tax rate, which decreased from 90% in the 1940s to 35% after the 1980s.