Most research on the social safety net focuses on understanding its effects on recipients and their families. We show that these effects extend beyond recipients’ families. Using a regression discontinuity design and administrative data, we study a Danish policy that drastically cut welfare benefits for refugees, leading to increased crime among the affected individuals. We link refugees to their neighbors and show large increases in crime among non-Danish neighbors. These spillover effects intensify over time and persist even when the direct effects on refugees’ crime stop growing. Due to spillovers, we estimate an increase of up to 23% in the marginal value of public funds associated with leaving welfare benefits to their pre-reform level. We explore several mechanisms for the spillovers, such as changes in employment opportunities, welfare use, and policing. We find evidence most consistent with peer effects in crime among young, unmarried peers from the same country of origin.