U.S. regions differ in some obvious ways: linguistics, culinary traditions, income distribution… In the two graphs, we show median family income (top) and the skewness of family income (bottom) for U.S. Census regions. Notice in the top graph that the South has remained persistently poorer than the rest of the regions, without much sign of convergence. In the beginning of the sample, in 1953, median income in the South was about $10,000 less than in all the other regions. In 2014, it still trails the Midwest and West by about as much. The Northeast’s median income, however, started its climb above all the other regions in the 1980s.
Compare this picture of between-region inequality with a picture of within-region inequality. In the bottom graph we look at the skewness of income, defined by the ratio of mean over median incomes. It is always greater than 1 because the wealthier top end of the distribution accounts for more of the variation than the poorer bottom end. The South was once the most “top heavy” region, with a more upwardly skewed distribution than any other. But it has since fallen back in line with the Northeast and West. However, the Midwest has remained consistently less upwardly skewed. This gap began to materialize significantly in the 1980s, just as the Northeast median earnings were beginning to pull away.
How these graphs were created: For the top graph, search for “real median family income in census region” and add the series for the West, Midwest, Northeast, and South to the graph. For the bottom graph, search for the same series and modify each one as follows: Once you’ve added the median series, use the “Add Data Series” / “Modify existing series” options to incorporate the corresponding mean series. Then use the “Create your own data transformation” option to apply the formula b/a. Repeat this for the three other regional series.
Suggested by David Wiczer