We assess the role of inequality, poverty and drug prohibition, in explaining homicide rates(HR) in Colombia using panel data at the municipality level between 1990 and 1998. We use maximum likelihood estimation to evaluate several specifications of spatial models. When we pool the data we find a significant relation between the HR and poverty which is reinforced when the inequality variables are included in the regression. We use rates of arrest related to drugs as proxy for drug prohibition enforcement, and find no evidence of any net effect in this exercise. We proceed to estimate the model for every year since 1991 to 1998, and find mixed results along the analyzed period, with statistically significant positive coefficients for some years, evidencing a positive net effect for those years of drug prohibition on violence. This result calls for caution in the implementation of global policies meant to improve specific countries’ welfare, such as drug prohibition policies. On the other hand, our proxy for income inequality is not related to violence, while its relationship with poverty is weakly positive.
