Regional income variation certainly seems significant for the single measure picked for the article. There's more detail in Derek Thompson's article [1] with ma

Regional income variation certainly seems significant for the single measure picked for the article. There's more detail in Derek Thompson's article [1] with maps for both "absolute upward mobility" and "relative mobility" measures. Even though it uses national income rankings as a base, the "relative mobility" measure compares outcomes between people growing up in the same commuting zones. That should exclude the affect of differing regional incomes. It's broadly similar, but more variation is visible in the South than in the other maps. Area income seems to offer a compelling explanation for one local result highlighted in the paper [2]: "San Francisco has substantially higher relative mobility than Chicago: y ̄100 − y ̄0 = 25.0 in San Francisco vs. y ̄100 − y ̄0 = 39.3 in Chicago. But part of the greater relative mobility in San Francisco comes from worse outcomes for children from high-income families. Below the 60th percentile, children in San Francisco have better outcomes than those in Chicago; above the 60th percentile, the reverse is true." Someone on a >60th national income percentile in San Francisco is going to be in a much lower local income percentile. That would make them relatively disadvantaged in comparison to someone on a >60th national income percentile in Chicago. And if they leave the city for somewhere with a lower cost of living, even if they stay in the same local income bracket as their parents they will appear to be doing worse in terms of national income ranking. [1] http://www.theatlantic.com/business/archive/2014/01/the-geography-of-the-american-dream/283308/ http://www.theatlantic.com/business/archive/2014/01/the-geog... [2] http://obs.rc.fas.harvard.edu/chetty/mobility_geo.pdf http://obs.rc.fas.harvard.edu/chetty/mobility_geo.pdf