See the Migrants, Cross the Street? Endogenous Selection Out of Exposure to Migrants

Using machine learning and natural language processing methods applied to novel big data of establishment reviews, and with a discrete choice model, I analyze whether the everyday consumption and mobility patterns of Polish natives shift with Ukrainian migrant arrival, away from local neighborhoods where the likelihood of migrant exposure is the highest, and the concentration of these effects in places which are otherwise most conducive to social mixing. As such, I consider how individual choices affect whether, and under which conditions, intergroup contact can take place, and hence how likely the potentially positive effects of contact are to be realized.

The study is conducted in the context of the immigration of more than one million Ukrainian refugees to Poland since the Russian full-scale invasion of Ukraine in February 2022.