Organized crime, violence and State response

The first chapter of this PhD thesis develops a model analysing the interplay between the Italian state and mafias. I show that targeting criminal profitability reduces conflicts within criminal organizations but increases retaliation against the state, while law enforcement decreases both. Using assassinations of public officers as the dependent variable, I support these findings by exploiting the introduction of Italy's asset forfeiture law and instrumenting confiscations with exogenous variations in US drug demand. This highlights three major epochs of post-WWII Italian history: the corrupted state, the enforcement-retaliation, and the sinking. Chapter two examines the effects of targeted repression of organized crime on inter-clan violence and illegal activity. We investigate the trade-off between curbing violence and illegal activity. If clans fight for territorial control, targeting the strongest reduces violence; however, if surviving clans are the most productive, it boosts illegal activity and profits. Targeting the weakest clans has opposite effects. If clans maintain a peaceful territory-splitting agreement, targeting the strongest may raise violence by triggering a succession war. Conversely, targeting the weakest may let the strongest thrive peacefully. Our theoretical analysis interprets the evolution of violence and illegal activity after different repressive policies in Italy, Mexico, and Colombia. Chapter three presents a model revealing that economically disadvantaged factions gain the most from warfare, even if outnumbered or outgunned. This is because they have more to gain and less to lose from the conflict. Targeting the most aggressive belligerents could paradoxically fuel new conflicts, as shifts in asymmetry may tempt rival factions to attack. I test these predictions using a database on militarized interstates disputes spanning the last two centuries and instrumenting countries' GDP per capita differences with exogenous variations in rainfall.