Conflict and Female Leadership: Evidence from Colombia

Abstract

This paper examines whether female leadership can reduce violence in the context of the Colombian conflict during the late 1990s and 2000s. To identify the effect, I leverage close mayoral races contested between a woman and a man. I find that after the narrow election of a female candidate, a municipality experienced a decline in the number of guerrilla attacks in the following electoral cycle amounting to 60% of the average incidence of guerrilla violence. I use a unique and novel dataset comprising the spatial distribution of guerrilla units and the gender of the command line to show that female leadership in the guerrillas is also correlated with lower levels of violence. Crucially, I show that whenever two female leaders coincide, violence decreases in a non-linear way. To understand the mechanisms underlying the main effects, I compile a novel dataset encompassing the campaign statements from a subset of mayoral candidates. Using unstructured text-analysis algorithms, I show that female mayors make use of more “peaceful” expressions, without diminishing their recognition of conflict. This suggests that female leaders have different “preferences” compared to their male counterparts. Finally, I combine administrative and survey data to show that there are no gender differences in terms of performance during office or political ideologies, and that the effect is not driven by the behavior of other conflict actors.