A friend sent me this interesting link to the “Political Calculations” blog in response to my recent post on distorted metrics of changing income inequality. The PC link is from about 10 months ago, but it is very timely nevertheless. It compares trends in three versions of the “gini coefficient” for the U.S., a common measure of inequality. A gini of zero indicates absolute equality of income across quantiles. A gini of one indicates that all income flows exclusively to the top quantile. The most interesting feature of this comparison is that the gini calculated for individual income earners has shown no trend up or down since about 1960, while the gini for households has trended upward since about 1970. This difference implies that the much-bemoaned increase in inequality is actually the result of changes in household structure, as opposed to earnings increasingly skewed toward elite individual earners.
In two follow-up posts (“The Widow’s Peak” and “The Men Who Weren’t There“), the author(s) identifies some demographic factors that were important in creating the upward trend in the household gini. Single-person households grew dramatically throughout this period, as did the number of seniors (aged 65+) living alone. The number of senior, female single-person households grew even more dramatically. Many of these women were either widows or never had good opportunities to marry because so many males in their age cohort were killed in World War II. Unfortunately, they constituted a group of very low-income households. There were other reasons for growth in the number of low-income, single-person households, such as increases in the divorce rate and perverse welfare-state incentives. The following lament by the author(s) in the last follow-up post is noteworthy:
“To us, it’s more remarkable that so many economists and politicians insist on focusing on the opposite end of the income spectrum in attempting to blame the highest income-earning Americans for that increase.”
Of course, the trends in ginis shown at the links above are subject to the same criticism made by the sabermatrician discussed in my earlier post: the comparisons over time implicitly assume that the quantiles used in the calculations are static, composed of the same sets of households or individuals, but they are not. Therefore, they tend to overstate trends toward greater inequality.
The fact that household structure has so much to do with trends measured by standard household gini coefficients, that the gini for individual earners has no trend, and that in any case, fixed quantile comparisons overstate trends in inequality, all suggest that redistributionist policies are misguided and unnecessary. And those kinds of policies tend to undermine the growth of the economy. Only policies designed to boost economic growth can help households across the income distribution to prosper. Moreover, unraveling the negative incentives built into current social programs would help to stabilize household and family structure.