There is *so much* variation even across Europe on which immigrants come in and how their children do. Figure (Boustan et al) shows rank gap vs. native-born for immigrants and their children. One of the richest graphs around. What is going on here? 1/

Why are immigrant kids so downwardly mobile in Austria? Holland—Daughters of immigrants converge to the mean in one generation, sons stay very poor. Why?? Highest mobility engines are France and Sweden. Immigrants do well in UK and USA, but arrive with more skills too. 2/

In Spain, Italy, and Canada, even the first-gen immigrants have higher income than the mean. Is this opportunity and mobility, or is it selection? I wish we could see the full time path, not just a cut of each generation. (We observe parents in 1994, children in 2014). 3/

Big differences in who is immigrating — in some countries they match the native-born distribution. In Aus, France, Norway, Denmark, Holland, Sweden, they are much poorer. 4/

I would be curious to see it broken down by sending region. It seems relevant for policy, immigrants from Western Europe, Eastern Europe, and LMICs feel like important different categories to me. 5/

The authors say that sending country doesn't explain much of the variation — we see here that there's high upward mobility in Canada and low in Austria, whether you come from Morocco, Yugoslavia, or Turkey. But it feels like something you'd want to hold constant throughout. 6/

It seems hard to separate place effects from selection effects — if Austria is very bad for immigrants, then even after controlling for education, it's probably the case that the set of high-education immigrants choosing Austria is different from those choosing Canada. 7/

So I'm not yet convinced these are causal effects of place. But the descriptive results are super interesting, and the starting point for asking why selection and outcomes are so different across places. 8/

I hope there is a big data release when this paper gets published (or sooner!) as it seems like a lot more to study here. Paper is Boustan et al. 2025, link here, List of authors: /N