Peace, Bullets, and Asymmetric Dividends: The Heterogeneous Effects of the 2016 FARC Peace Agreement on Municipal Homicides in Colombia

Abstract Does a peace agreement reduce lethal violence evenly across a country, or only where the war was worst? This article estimates the heterogeneous effect of Colombia's 2016 peace agreement with the FARC insurgency on municipal homicide rates, using a continuous difference-in-differences design across all 1,079 Colombian municipalities over the years 2013 to 2023. The treatment is a municipal pre-violence index: each municipality's average homicide rate in the three years before the agreement, divided by the national maximum, which serves as a proxy for exposure to the conflict. The index is validated as the first stage of an instrumental-variables design, in which exposure to documented guerrilla conflict events, population, and region jointly predict it. The two-way fixed-effects estimate implies that the agreement reduced homicides sharply and in proportion to prior violence: the implied reduction rises from about 6 percent at the twenty-fifth percentile of the index to about 44 percent at the ninety-fifth. An instrumental-variables specification that isolates the component of exposure predicted by prior guerrilla activity yields a smaller but still negative and significant effect, consistent with attenuation from measurement error. Event-study estimates confirm parallel pre-agreement trends and a sharp break in the year of the agreement that persists through 2023, and a permutation test that randomly reallocates the index across municipalities rejects the estimate as a product of chance. The peace dividend in homicides was therefore real but steeply graded: concentrated in the municipalities that had suffered most, and close to negligible elsewhere. Post-conflict security policy, the findings imply, should be calibrated to the prior geography of the war rather than distributed uniformly.