Paul Krugman blames sprawl the lack of upward mobility among children raised in households in the bottom of the income distribution. I have often argued that density can facilitate upward mobility, as it implies greater access to employment opportunities within a given commuter-shed. But there are a few problems with Krugman’s analysis.
First, Krugman draws on William Julius Wilson to suggest that deindustrialization in central cities is a key part of the upward mobility story:
A quarter-century ago Mr. Wilson, a distinguished sociologist, famously argued that the postwar movement of employment out of city centers to the suburbs dealt African-American families, concentrated in those city centers, a heavy blow, removing economic opportunity just as the civil rights movement was finally ending explicit discrimination. And he further argued that social phenomena such as the prevalence of single mothers, often cited as causes of lagging black performance, were actually effects — that is, the family was being undermined by the absence of good jobs.
The evaporation of a certain set of economic opportunities for less-skilled workers in city centers was indeed a heavy blow. The more difficult question to answer is why we see such a sharp divergence in outcomes between women and men, an issue David Autor and Melanie Wasserman explore in “Wayward Sons.” And though Autor and Wasserman don’t pretend to offer conclusive answers, they acknowledge the possibility that the direction of causality might run both ways, i.e., contra Krugman, it’s not simply that the absence of good jobs led to the prevalence of single-headed households (though this dynamic did indeed play a role), but that the prevalence of single-headed households seems to have contributed to male underperformance. Moreover, Lawrence Mead has observed that male non-work has persisted even during periods of tight labor markets. In Expanding Work Programs for Poor Men, Mead noted that even the most generous voluntary work programs, like the New Hope Project in Milwaukee, which offered generous benefits in return for working thirty hours a week, found it very difficult to attract jobless men.
Second, Krugman arguably underemphasizes the spatial dimension of upward mobility while overemphasizing the existing level of inequality:
When the researchers looked for factors that correlate with low or high social mobility, they found, perhaps surprisingly, little direct role for race, one obvious candidate. They did find a significant correlation with the existing level of inequality: “areas with a smaller middle class had lower rates of upward mobility.” This matches what we find in international comparisons, where relatively equal societies like Sweden have much higher mobility than highly unequal America.
One of the reasons why relatively equal societies might have higher mobility than relatively unequal societies is that relatively equal societies by definition have more compressed income distributions. That is, if households in the bottom fifth in Societies A and B have roughly the same income, yet households in the top fifth have much higher incomes in Society B than in Society A, it is not unreasonable to assume that relative mobility across household income quintiles will be higher in Society A than Society B, as there is less distance to travel between the bottom and the top. Also, Krugman cites the work of Miles Corak to make his point about international comparisons, and Scott Winship has written extensively about the limits of Corak’s analysis.
The study Krugman cites found that several U.S. metropolitan areas had levels of upward mobility for the poor comparable to the international leaders. See the discussion by David Leonhardt:
[T]he parts of this country with the highest mobility rates — like Pittsburgh, Seattle and Salt Lake City — have rates roughly as high as those in Denmark and Norway, two countries at the top of the international mobility rankings. In areas like Atlanta and Memphis, by comparison, upward mobility appears to be substantially lower than in any other rich country, Mr. Chetty said.
Especially intriguing is the fact that children who moved at a young age from a low-mobility area to a high-mobility area did almost as well as those who spent their entire childhoods in a higher-mobility area. But children who moved as teenagers did less well.
Yet these U.S. metropolitan areas (Seattle, Salt Lake City, Pittsburgh) have much higher Gini coefficients than Denmark and Norway. By way of comparison, the Gini coefficient (0 is perfect equality, 1 is perfect inequality) for the U.S. is .467, it is .248 for Denmark and .250 for Norway. Seattle has a Gini coefficient of .439, Salt Lake City has a Gini coefficient of .417, and Pittsburgh has a Gini coefficient of .459. If there is a tight association between inequality and mobility, how is it that Seattle and Salt Lake City and Pittsburgh are roughly matching the upward mobility performance of Denmark and Norway with levels of inequality that are subtantially higher? Again, this doesn’t mean that inequality is irrelevant. But if Pittsburgh (.459) and Denmark (.248) are in roughly the same ballpark, it seems that we ought to pay close attention to what Pittsburgh and cities like it are getting right.
The Gini coefficient for Atlanta is .452 while the Gini coefficient for Memphis is 0.478. Memphis clearly has an extremely high Gini coefficient. Note, however, that Atlanta, a metropolitan area that fares very poorly in terms of upward mobility for the poor, actually has a lower Gini coefficient than Pittsburgh, a metropolitan area that fares extremely well. Several other large U.S. metropolitan areas with higher Gini coefficients than Atlanta also far surpass Atlanta in terms of upward mobility for the poor, including highly unequal San Francisco, New York, and Los Angeles.
I want to stress than the correlation between low levels of inequality and high levels of mobility makes perfect sense, among other things because it is easier to climb a shorter income ladder than a taller one. The irony, however, is that an emphasis on inequality might distract us from a far more important issue, which is integration. Atlanta and Pittsburgh demonstrate that two metropolitan areas with similar levels of inequality can produce markedly different results for poor households. If one metropolitan area is highly segregated, with poor people living very far from rich people, while another is relatively integrated, with poor people and rich people living in close proximity, or with good transit links connecting the two groups, it seems pretty clear that poor people are much better off in the latter metropolitan area.
Rather than endlessly recapitulate the inequality debate, we need to have a more robust debate about economic geography. Progressive taxation gets all of the attention. But local land-use regulation that limits density, and that make it more difficult for poor workers to live within easy commuting distance of rich customers, gets virtually none.