Arming opposition: Measuring the effect of arms transfers on internal conflict

The main contribution of this paper is to provide the first causal estimation of the effect of weapons imports on confl icts in the recipient country for a global set of suppliers. The second main contribution is to suggest a novel approach to correct the errors with a Shift-Share (Bartik) instrument using arbitrary clustering and its implementation in STATA using the command acreg. Using small arms and light weapons country-year transfers data from 1992 to 2011, I estimate the effect of arms import on the con flict life-cycle (onset, duration, intensity and termination) and number of refugees from the recipient country. To address endogeneity issues, I am using supply shortages generated by suppliers' war involvement outside the recipient country's continent (Shift-Share instrumental variable) as exogenous negatives shocks on arms import. The two-stage least squares estimation shows that arms imports increase the onset of internal con flict, the number of internal con flicts, the percentage of battle-related deaths and the number of refugees fleeing the country. The 2SLS model predicts as well that if Europe would stop sending weapons to Africa for a year, it would reduce the number of refugees by 500,000.