Fig 1.
Region of Study migration passages and smuggling hubs (a) and area used for calculating sea conditions (b). Map of smuggling routes is recreated based on information from the IOM [24]. Sea condition zone is authors’ projection. Data on sea conditions is from the Copernicus marine environment monitoring service. Boundary files are from GADM [25].
Table 1.
Riots, sea conditions, and migrant flows to Italy.
Fig 2.
Trends in death rates and Sea conditions during sample period.
Daily death rates are calculated using IOM data on migrant deaths and disappearances in the Mediterranean and UNHCR intake data for the Italian coast. Wave heights are drawn from the Copernicus platform.
Fig 3.
Impact of Italian intervention.
(a) Weekly time series of actual (solid) and predicted (dashed) arrivals. Blue vertical bar indicates the week of the intervention. As expected, fit statistics suggest the predicted variation maps well to the pre-treatment period and poorly to the post-treatment period. (b) Regression-based estimates of the reduction in migrant flows following the intervention, relative the predicted flows from the prior year. Reduction is equivalent to 0.6 standard deviations. Outcome mean is -387.1; standard deviation is 595.2. 95% confidence intervals shown. (c) Monthly trends in migrant flows to three receiving countries. Notice large reduction corresponds to minor displacement through Spain and Greece. (d) Monthly trends in arrivals and deaths. Note strong correspondence between two trends (R2 = .79).