Asylum-Seeker Reception and Crime: the Case of Italian Dispersal Policy

Italy's position on the Central Mediterranean route made it the primary point of sea entry into the European Union. From 2014, the Italian government dispersed asylum seekers across the country through Temporary Reception Centers (CAS). These centers supplied food and accommodation, but not the language, legal, and labor market support of the municipal SPRAR network. In the Becker-Ehrlich framework, the supply of offenses depends on the legal earnings that an hour of crime costs the offender. A reception system of this kind is therefore expected to raise local offending. I test this prediction on a panel of 103 harmonized Italian provinces, observed between 2014 and 2019. I measure exposure as the number of asylum-seekers hosted in Hotspots, CPA and CAS reception centers over the resident population, and I match it to the offenses that the police reported to the judicial authority in the following year. I estimate a two-way fixed effects model with five time-varying controls, so the effect comes from the change in the share that each province hosted over time. I test the exogeneity of the treatment with a balance test on pre-crisis economic conditions and with the crime trends of the eight years before the CAS system began to fill. I find no significant effect. The coefficient on total crime is indistinguishable from zero, none of the four macro-categories differs from zero, and across fifteen alternative specifications the coefficient on total crime always lies within [-0.072,\,0.090]. Among the twenty-six individual offenses, five are significant at the 10 percent level, three of them with a negative sign. The negative correlation across provinces reflects where the reception system placed asylum seekers, not the effect of their presence. Since crime data are available by province, my estimates leave open whether the response differs across municipalities. The increase in crime feared from the dispersal does not appear in the provincial record, while previous research has already measured the political reaction to the same policy.