Someone pointed me to this article by David Leonhardt, Amanda Cox and Claire Cain Miller, who write:

Based on the earnings records of millions of families that moved with children, [a new study] finds that poor children who grow up in some cities and towns have sharply better odds of escaping poverty than similar poor children elsewhere. . . .

The findings suggest that geography does not merely separate rich from poor but also plays a large role in determining which poor children achieve the so-called American dream. . . .

“The data shows we can do something about upward mobility,” said [Raj] Chetty, a Harvard professor, who conducted the main study along with Nathaniel Hendren, also a Harvard economist. “Every extra year of childhood spent in a better neighborhood seems to matter.”

The article is excellent — it’s a sober, data-filled discussion of an important project, a model of how social science and journalism can work together. And the accompanying graphic is clear and informative as well.

I just have one criticism, which is the occasionally casual treatment of causality. The statement on the graph is clear: “How much extra money a county causes children in poor families to make, compared with children in poor families nationwide.” That’s a strong statement, but the news article makes the convincing case that the research project, by specifically looking at people who moved in childhood, can give some leverage on the causal claim. So that’s fine.

But then there’s this:

Previous estimates suggested that New York City was a good place for lower-income children to grow up: Children raised in lower-income families in New York had above-average outcomes in adulthood.

But New York appeared above average in part because it has a large number of immigrants, who have good rates of upward mobility no matter where they live: Nothing about New York in particular caused these children to do better.

That doesn’t make sense to me: “Nothing about New York in particular” caused these children of immigrants to do better? But maybe there’s a reason all these immigrants came to New York, no? I understand the rationale for performing an analysis that compares like with like, but it also seems to me that upward mobility of immigrants is a big part of the American economic story, and by subtracting this out, you’re removing something important.