Migration Through Multiple Lenses: Policy, Distance, and Discrimination

This thesis advances the empirical understanding of migration across three dimensions: the consequences of EU border ex- ternalisation on migrant safety; the determinants of migration distance in West Africa; and whether terrorism shapes ethnic disparities in policing. Each study combines high resolution data with causal and machine learning methods. The first study examines the EU–Turkey Statement of March 2016 us- ing a Matrix Completion estimator on monthly Frontex data. It finds a significant diversion from the Eastern to the Cen- tral Mediterranean, approximately 2,000 additional crossings between April and December 2016, and estimates 45 ± 3 ad- ditional deaths attributable to the agreement, showing that single-route containment policies are sub-efficient. The sec- ond study uses over 60,000 IOM observations from eight West African countries to predict intended migration distance via Tree models. Results reveal a bimodal distribution; employ- ment status, nationality, and GDP are the strongest predictors. Conflict-induced mobility is temporally volatile, particularly in 2023 as insecurity intensified across the Sahel. The third study investigates whether Islamist attacks exacerbate ethnic disparities in policing using daily stop-and-search data from England surrounding six attacks between 2017 and 2020. A stacked difference-in-differences design finds no evidence that attacks increased stops directed at ethnic minorities relative to White individuals, suggesting frontline policing is resilient to short-run surges in threat perception. These three studies trace migration from the policies shaping transit routes, to the forces determining how far people travel, to the institutional responses minorities face at destination.