"Channels of Peer Effects and Guilt Aversion in Crime: Experimental and Empirical Evidence from Bangladesh"

　　 This study empirically disentangles the channels of peer effects in crime through an experiment conducted in rural Bangladesh. The first part of this study assumes that individuals exhibit guilt aversion, which predicts the peer effects via guilt sensitivity and belief. By incorporating peer effects in a take-away game, the criminal player is informed about the victim player's belief only in the treatment group, so that the peer effects in the treatment group are driven only through guilt sensitivity. The experime ntal results suggest that peer effects affect and bring about changes in belief. The second part elicits guilt sensitivity to test guilt aversion. I find robust supporting evidence for my results, and reject the alternatives such as pure altruism and trustw orthiness. Finally, external validity is also confirmed: the criminal behaviour of subjects in the experiment is correlated with their attitude towards illegal activities in the real world, and individuals are less likely to suffer from property crime in villages with a higher guilt sensitivity neighbourhood.