Some behavioral legacies of exposure to gangs, riots, and war

Abstract

The dissertation is composed of three stand-alone chapters that examine whether exposure to violence has consequences for prosocial behaviors. The first chapter investigates interethnic cooperation after ethnic conflict in southern Kyrgyzstan. I study the aftermath of the 2010 Osh riots, which saw Kyrgyz from outside the city kill over 400 Uzbeks. I implement a preregistered representative survey including incentivized measures of prosocial behavior. My causal identification strategy exploits variation in the distance between Uzbek neighborhoods and the barracks where armored military vehicles were stored. These vehicles - seized by the Kyrgyz attackers during the riot - were instrumental in orchestrating the violence. The results demonstrate that residents of damaged neighborhoods show substantially lower levels of prosocial behavior. Importantly, using a within-subjects design, I show that the reduction is the same for prosocial behavior toward in and outgroup members (coethnics and non-coethnics, respectively). I discuss several potential mechanisms linking destruction to reduced prosociality, pointing to social disintegration as the most likely channel. The second chapter investigates the effect of civil war on community cohesion in rural Colombia. I compare communities exposed to military conflict between symmetric and asymmetric competitors. While symmetric war is fought with relatively conventional means between clearly defined camps, in asymmetric war a weaker side adopts guerrilla tactics and boundaries between camps become blurred. I compare community cohesion in settlements exposed to these two types of conflict using survey data from 224 Colombian villages. The results show that symmetric war increases community cohesion measured as participation in community organizations. By contrast, asymmetric war has a detrimental impact on trusting behaviors. These results are consistent with the conjecture that asymmetric war fuels denunciation-driven violence of civilians, which creates unobservable community fractures that make individuals more prudent in relations with their neighbors. The third chapter examines the effect of gang exposure at school on youth’s social behaviors and attitudes. To identify the effect of gang exposure, I exploit plausibly exogenous variation in exposure to gang-affiliated classmates across schools in rural Colombia. Specifically, I exploit the fact that schools in rural Colombia are relatively unsegregated with respect to students’ socioeconomic backgrounds. Moreover, the distribution of youth gangs across these schools is plausibly orthogonal to the studied outcomes. The analysis of survey data from rural Colombia - subsequently replicated on data from the United States - reveals gender differences in behavioral adjustments to youth gangs. I find that girls react to male gangaffiliated classmates by intensified involvement in prosocial organizations. Boys, by contrast, adjust to male youth gangs by adopting more antisocial attitudes. These patterns are reversed in the case of exposure to female youth gangs.