タイトル：Guilt Aversion and Peer Information in Crime: Evidence from Experiment and Survey Data in a Developing Country
It has been discussed that the existence of peer effect causes the disparity in criminal incidence, but few studies examine how it occurs. I conduct an artefactual field experiment in rural Bangladesh with randomly sampled participants to identify what causes the peer effects. Particularly, I test two channels based on the guilt aversion preference; through the change in guilt sensitivity and the second order belief. A novel contribution of this experiment is that it elicits the guilt sensitivity at the individual level. I find that the patterns of criminal behavior are consistent with the guilt aversion rather than the altruism and trustworthiness; participants with higher guilt sensitivity and/or lower second order belief are less likely to commit crime. The peer effect occurs through the changes in the second order belief; when crime is common, individuals anticipate that the others expect higher risk of crime victimization, which in turn declines the guilt from committing crime. By using the survey data collected from the same participant households, I show the validity of the elicited guilt sensitivity; individuals are less likely to experience crime victimization in the villages where villagers have higher guilt sensitivity.