As Western societies diversify, a pressing question is how to facilitate the integration of newcomers. In policy circles and in academia, a popular solution is to provide opportunities for frequent interaction between newcomers and natives places like schools and neighbourhoods. As it stands, causal evidence of the effect of frequent interaction on various measures of integration (interethnic friendships, political participation, institutional trust) is scant, especially when it comes to longer-term effects. In addition, integration may be helped by positive contextual conditions in schools (i.e. teachers' ethnic background, civics courses) and neighbourhoods (i.e. the presence of public services, the presence of meeting places).

In this paper, I study integration dynamics in schools and neighbourhoods. For neighbourhoods, I use panel registry data of the whole Dutch population linked to panel survey data over a period of fifteen years and to an original survey. I study whether changes in neighbourhood composition affect integration, looking both at neighbourhood changes and at individuals who move. Moreover, I leverage quasi-random assignment of neighbours to study the causal effect of neighbour composition on measures of integration. Finally, I look at changes in the quality of public services, and the presence of meeting places.

Then, I use social network data from schools in Germany, Sweden and the Netherlands to study whether quasi-random assignment to ethnically diverse classrooms fosters friendship formation across ethnic divides. I also look at its long-term effects, and at the effect of teacher ethnicity and changes in civics curricula.