The Brookings Institute study is a clever bit of work. It uses data from the 2014 and 2016 American Community Survey data on household income distribution, for the 100 largest metropolitan areas (Metropolitan Statistical Areas, or MSAs), and their principal (largest) cities.
The measure of “income inequality” in the Brookings study is the ratio of the area’s household income at the 95th percentile (only 5% of the households make more than the 95th), to the household income at the area’s 20th percentile (20% of the household make less than that.)
This data is available from Table B19080 in the Census Bureau’s American FactFinder (AFF). What makes the Brookings study an extra chore (for them, at least) is that the Census Bureau puts a $250,000 cap on the 95th percentile for all data reported in the standard AFF tables. So, the Brookings people examined data from the 2016 ACS Public Use Microdata Sample (PUMS) to calculate a better estimate for about a dozen or so metro areas and their principal cities. That seems like a lot of work!
What puzzles me is why the Brookings Institute researchers didn’t use the most direct index of income inequality in use in the world, and as provided by the Census Bureau, the Gini Coefficient?