The Chetty et al. paper on inter-generational mobility has been getting a lot of attention. Social mobility data are essential to any discussion of economic performance -- and a welcome antidote to all of the misleading conclusions drawn by commentators who simply compare cross-sections. (You know who you are.) People move between quintiles and it is most informative if we find a way to track their progress. The Chetty et al. study develops a clever way to use IRS data to make comparisons of incomes, between parents and their offspring. Have offspring moved up after 30 years? To their great credit, the authors say that explanations for the geographic variation in their results remain to be explained.
The authors did include a bunch of bi-variate correlations which they point out are only suggestive of where people experience the most and least intergenerational mobility. The usual suspects pounced on these and concluded it's the "sprawl" that inhibits progress. I have often posted my thoughts on this misunderstanding. Here is my most recent.
Besides, "sprawl" is a vague and pejorative use. Suburbanization occurs everywhere and is much too varied to be described in overly simple terms.
Wilson and Singer of Brookings report where international immigrants to the U.S. choose to settle. The authors do this for the 100 largest U.S. metropolitan areas for 2000 and 2010. For each of these, they note whether or not the foreign born end up in the suburbs. For the whole set of metros (slightly fewer than 100 in 2010), 56.1% of the foreign born settled in the suburbs in 2000 while 60.6% did so in 2010.
I point this out because these immigrants, by definition, are not in the Chetty at al. data; they were not here (not in the IRS files) in the first year of the study. And by virtue of coming here, they are among the most upwardly mobile (see Lant Pritchett interview).
In the fast-and-loose manner that some have digested the Chetty et al. study, we could conclude that sprawl causes upward mobility. What the hell!
David King's analysis on point.