It's well known at this point that not only does the United States have an unusually large level of income inequality compared to other rich countries, but we have a low level of intergenerational mobility. Kids whose parents have low incomes, in other words, are very likely to themselves grow up to have low incomes. But the United States is also very large compared to most other rich countries, which raises the question of how uniform that pattern of mobility is. A new study exploiting a richer dataset than has previously been available shows that there's quite a bit of variation. In diverse set of metropolitan areas including most of the major coastal cities but also Salt Lake City and most of the low-population communities of the Great Plains, kids born into the bottom fifth of the national income distribution have a fairly high chance of moving up into the top fifth. Conversely, there's a broad crescent across most of the American south and then a large pocket in the industrial midwest where intergenerational mobility is tiny.
You can see a version of the map from the authors' website here or play around with the extensive interactive tools that company David Leonhardt's excellent writeup here. Obviously when simply eyeballing the map you have to apply some common sense to the fact that that the large high-mobility area on the Plains contains many fewer people than the low-mobility area in the southeast.
So what drives this? The study does not really make a high-powered effort to draw strong causal inferences. But the study does show that kids who moved into a high-mobility area at a young age do about as well as the kids born in high-mobility areas, but kids who move as teenagers don't. So there seems to be a factor that isn't parent-driven. The authors report that tax policy, the existence and affordability of local colleges, and the level of extreme local wealth do not appear to be strong correlates of intergenerational mobility. Metro areas where the poor are geographically isolated from the middle class have less intergenerational mobility, while metro areas with more two-parents households, better elementary and high schools, and more "civic engagement" (measured through membership in religious and community groups) have more.
This is groundbreaking empirical work, though I'm not sure it revolutionizes my thinking about public policy beyond suggesting that we perhaps shouldn't pay too much attention to the bowling for fascism problem. There is a current of thinking on the left in the United States that holds that people care too much about school quality, and I think we can add this study to the (long) pile of reasons for while the conventional wisdom is correct about this and actually school quality is quite important. The merits of mixed-income neighborhoods strengthen the case against zoning out the poor with minimum lot size requirements, rules against apartment buildings, and trailer park bans but of course that's all stuff I already believed.