The spread of "sprawl" as latest explanation of black poverty

In the New York Times, David Leonhardt ponders critics (like me) pointing out that Raj Chetty's map of upward mobility (defined as chance child in bottom 20% of income in 1996 makes it to top 20% in income as a younger adult) in the United States looks like a map of the Where the Blacks Aren't. He answers by saying Chetty proved that wasn't true.

The simplest way to explain their conclusion may be to point out that upward mobility tends to be rare for both blacks and whites, as well as for Latinos, in low-mobility areas. In Charlotte, Atlanta and Indianapolis, low-income white children have also tended to grow up to be low-income adults. 

To help demonstrate this pattern, the four researchers – Raj Chetty and Nathaniel Hendren of Harvard and Patrick Kline and Emmanuel Saez of the University of California, Berkeley – have produced another map, showing mobility only for ZIP codes that are at least 80 percent white. (ZIP codes that are less than 80 percent simply appear as blank on the map.)

But the message is clear: the mobility patterns look overwhelmingly similar in this map and in the map above showing all metropolitan areas. 

It’s worth pointing out that race may still play a role in creating these patterns. “Racial shares in an area do matter,” Mr. Chetty says. “But it’s not race at the individual level. It’s race at the level of the ‘commuting zone,’” he added, using the researchers’ term for a region. Whatever the differences are between high-mobility and low-mobility regions, they seem to apply to residents of every race. 

No, race at the individual level still very much matters when you are looking at the chances of the bottom 20%. In the last map, these Southeastern and Rustbelt districts might be ones where blacks make up less than 20% of the population, but they still make up a much larger fraction of the bottom 20% than in the Great Plains.

No, a big part of this is simply Galtonian regression toward the mean. It shouldn't be controversial that whites have higher mean incomes than blacks. So, white children who find themselves growing up in the bottom 20% regress part way toward a higher mean than black children growing up in the bottom 20%.  

Then Leonhardt trots out the liberal explanation du jour for the poor economic performance of heavily black metropolitan areas: sprawl.

Writing about the new study for The Atlantic, Matthew O’Brien laid out a specific case for how race might have created economic segregation in sprawl-filled regions: 

Atlanta, of course, is the prototypical case here: going back to the 1970s, it’s under-invested in public transit,

Okay, but on net African-Americans have been moving from public transit rich NYC to the sprawling Atlanta metropolitan area in pursuit of a better standard of living. Are blacks irrational about this? Or is it simply that Chetty's study inadequately controls for the very different costs of housing across the country?

Sprawl might be one factor, but how long does it take to come up with a list of old fashioned cities like Baltimore, Cleveland, St. Louis, Newark, Hartford, and Milwaukee with intense black poverty? (Of course, if, say, whites in Hartford flee to the suburbs, leaving a black core, then that's "job sprawl." That's so unfalsifiable that if Sir Karl Popper were alive today, he'd be spinning in his grave.)

Also consider the changing impact of local climate, which has only disparate impact effect on race: Think of the wage premium needed to attract workers due to the climate, with Honolulu at zero extra dollars and Point Barrow, Alaska at a bundle. There has been a big shift over the generations. They've had central heating in the Dakotas for a long time now, but air conditioning has spread in Georgia only in the second half of the 20th Century. The spread of AC means Georgia is filling up with workers newly happy about the climate, keeping wages down as it lowers the wage premium in Georgia relative to the Dakotas. In other words, the standard of living has gone up in Georgia due to summers becoming more tolerable. All else being equal, an endogenous change in technology that makes the climate seem better will lower the wage premium.