Does per capita income cause homicide rates? An application of an IV spatial model

Criminality is an increasing problem in urban centre. In this paper, we are contributing for this literature estimating the casual relationship betweenper capitaincomes on homicides rates. The simultaneity between those two variables and spatial spillover make our aim challenger. To deal with this problem, we are combine two regression technics: Instrumental variable and spatial econometrics. Using rainfall as an instrument, we have found that increase the GDPper capitain 1% reduces the municipality homicides in 0.79% in total, 0.5% directly and 0.3% indirectly. In other words, the indirect effect arises because of the influence of each variable on the neighbours' dependent variable and these spillovers amplify the GDPper capitaeffect on homicides rates. Another finding relevant is climate change. We have done a forecasting exercise decreasing the rainfall around 5 SE (- 28.5ml(3)) , then we have found that homicides rates increase 4%.