Too many academics play fast and loose with scale. I've seen the geographic unit of analysis change within one sentence.
A recent example in Bloomberg:
In a new working paper, Shoag and Peter Ganong, a doctoral student in economics at Harvard, offer an explanation: The key to convergence was never just mobile capital. It was also mobile labor. But the promise of a better life that once drew people of all backgrounds to rich places such as New York and California now applies only to an educated elite -- because rich places have made housing prohibitively expensive. (Shoag and Ganong visualized these changes in a series of excellent animated graphics.)
The states with the highest incomes also used to have the fastest-growing populations, as Americans moved to the places where they could earn the most money. Over time, that movement narrowed geographic income differences. In 1940, per-capita income in Connecticut was more than four times that in Mississippi. By 1980, Connecticut was still much richer, but the difference was only 76 percent. In the two decades after World War II, Shoag and Ganong find, migration explains about a third of the convergence of average incomes across states.
But migration patterns changed after 1980. “Instead of moving to rich places, like San Francisco or New York or Boston, the population growth is happening in mid-range places like Phoenix or Florida,” Shoag says. Lower-skilled people, defined as those with less than 16 years of education, are actually moving away from high-income states.
The offense is in bold (emphasis added). If you analyze interstate migration, then your conclusions should reflect that. Comparing "Boston" to "Phoenix or Florida" is a mistake. The unit of analysis in the working paper is the state (e.g. Florida). Leave cities and metros out of the discussion. Stick to states.
That grievance aired, "Why Has Regional Convergence in the U.S. Stopped?" is a fascinating read. People, regardless of income, used to follow the money. During the great migration, poor from the rural South streamed to the urban North. (Regional convergence) Now, people with less than a college degree are moving from high income states to low income states. (Regional divergence)
Enrico Moretti makes this very point in his book, "The New Geography of Jobs." In a recent interview with the American Enterprise Institute,
a controversy arises between Moretti and Shoag:
Differences in geographical mobility, coupled with increasing polarization among American cities, exacerbate income differences across education groups. Indeed, if the less educated people were more able and willing to move to cities with better job opportunities, the gap between college graduates and high school graduates would shrink.
Moretti thinks the unemployed need help moving to
where the rent is too damn high.
Shoag thinks the crux of the problem is that the rent is too damn high:
These results suggest that rising share BA in high BA areas ((Glaeser and Saiz, 2004) and (Moretti, 2012)) may be the result of out-migration by low-skill workers or domestic human capital production rather than increasing in-migration by high-skill workers.
Emphasis added. That's from footnote #31 on page 19. High housing prices are pushing out low-skill workers (Shoag) as apposed to high-skill workers agglomerating in a select few places (Moretti). But that's a huge "may be" that the paper can't answer.
Of course, that doesn't deter Virginia Postrel:
Finally, there’s the never-mentioned possibility: that the best-educated, most-affluent, most politically influential Americans like this result. They may wring their hands over inequality, but in everyday life they see segregation as a feature, not a bug. It keeps out fat people with bad taste. Paul Krugman may wax nostalgic about a childhood spent in the suburbs where plumbers and middle managers lived side by side. But I doubt that many of his fervent fans would really want to live there. If so, they might try Texas.
Postrel sucks at geography. I recommend Moretti over Shoag. Again,
Moretti from the interview with AEI:
One of the most striking facts about our society is how different the economic fortunes of cities can be. In the United States, for example, the wage of workers varies enormously depending on the city. The hourly wage of workers in a city like San Jose is more than double the hourly wage of workers in a city like Visalia. These cities are in the same state, are just 200 miles apart, and they share the same legal system, language, and culture—and yet from the point of view of their economies, they look like they belong to different continents. And the gap is not going away—it has been increasing for 30 years. It is not that San Jose has better natural resources or better transportation infrastructure than Visalia, or that its workers work longer hours. The entire productive ecosystem is different. It is hard to understand the reasons that make local economies so different, unless you start thinking about forces of agglomeration.
Emphasis added. That's an economist thinking like a geographer.
Jared Diamond would applaud Moretti's natural experiment. How would Shoag's analysis hold up using a similar research design? From the work I've done, I think Moretti is right and Shoag is wrong.