Do refugees cause crime?

Abstract

• There is not any statistically significant impact of the Syrian refugees on crime rates in Turkey in the short-run (1-year after the treatment). • Staggered difference-in-differences analysis shows that the provinces where refugees make up over 1 percent of the total population do not demonstrate any statistically significant impact on their crime rates in the long run, either. • There is a negative impact of treatment intensity on crime rates per capita when the impact of refugee population share is analyzed using an instrumental variables strategy. The same analysis for the crime rates per native resident produces null effects. • Overall, the empirical analyses provide causal evidence for a rather lower propensity of crime commitment among the Syrian refugees in Turkey compared to the country’s native population. The null effects of refugees on crime rates per native resident, on the other hand, can be interpreted as a non-significant spillover effect of refugees on native criminality. • Higher expected costs of crime commitment for refugees such as deportation and negligible adverse impacts of them, on average, on the natives’ labor market outcomes, among other mechanisms, might have played an important role in these findings. The impact of immigration on crime continues to stir heated debates in public policy circles around the world. Surveys indicate that host societies favor controversial measures because they are concerned about what they perceive to be an impingement exacted on their security with each new wave of migration. Seeking whether there is any truth to such perceptions, this paper analyzes the refugees’ impact on crime rates, using the case of Turkey which has started to host the world’s largest refugee population within any national borders due to the Syrian civil war. In doing so, the paper employs instrumental variables, difference-in-differences (DiD), and staggered DiD methods to explain if the conflict-fleeing Syrians have pushed Turkey’s crime rates higher in the short and the long run. It also controls for a multitude of time-varying provincial characteristics and presents a battery of robustness checks against various identification threats. As a result, DiD estimates show that refugees do not have any causal effect on the crime rates in Turkey. More strikingly, its IV estimates provide evidence for a rather negative effect on the crime rates per capita whilst finding a null effect on the crime rates per native resident in particular.