Seminario CEDE - Francisco Eslava

This paper examines whether female leadership can reduce violence in the context of the Colombian conflict. To identify the effect, I leverage closely contested local elections between male and female candidates, alongside a novel dataset with the location and gender of guerrilla unit commanders. I find that female leadership has a negative effect on conflict violence that amounts to a 60% reduction in the most demanding specification. I document even larger reductions in violence when both a female mayor and a female guerrilla commander are present, revealing that synergies between female leaders on opposing sides of the conflict further contribute to de-escalation. To understand the reasons behind this reduction in violence, I use text analysis on a novel dataset of campaign statements, revealing that female mayors employ less confrontational language when addressing the conflict. Additionally, I leverage timing of conflict events to show that guerrilla units with female commanders are less likely to retaliate following unilateral attacks. I interpret these patterns as evidence of a genderbased preference for de-escalation. Finally, to study how these preferences translate into lower violence, I build a simple theoretical model to show that gender differences in “negotiation skills” can account for the observed empirical patterns. Testing the model’s predictions and ruling out alternative explanations, I conclude that female leaders consistently opt for de-escalation over violent confrontation whenever possible.