This paper studies the socio-economic determinants of violence in the seven most important cities in Colombia. Derived from theories rooted in classic works, including Becker's paradigm, criminal inertia, social disorganisation, urbanism and strain, we formulate several hypotheses on the socio-economic determinants of violence. To test these hypotheses, a dynamic panel data analysis is employed. The analysis shows that cities’ deprivation and high population density are strong predictors of homicide rates. Comparing the results among Colombian cities, we find support indicating that city-level homicide rates are influenced by the city's level of development. Moreover, we find evidence that economic growth, inequality, poverty and human capital influence violence in the cities studied, which could generate negative effects on the economic and social development of Colombia.
That is from a new paper by Alexander Cotte Poveda (Journal of International Development, October 2012). A draft is here.
Probably the analysis should have added governance or efficiency of the judiciary (or a proxy of them) as explanatory variable.
The author explains:
Another finding consistent with our hypothesis (Hypothesis 2) concerns drug seizures, measured as marijuana, cocaine, and coca seizures, where the relationship between violence and drug sized is negative and significant for marijuana seizures, indicating that a higher level of drug seizures generates a lower motivation to commit an illegal activity. However, Becker et al. (2004) have demonstrated that drug trafficking as an illegal activity for suppliers and consumers is a not efficient strategy to drug control, whereas legalizing drug use and taxing consumption may be more effective than continuing to prohibit the use of drugs.