Eliciting Guilt Sensitivity to Predict Real-World Behavior

This study tests guilt aversion by experimentally eliciting guilt sensitivity of villagers in Bangladesh and evaluating its impact on real-world behavior. In a trust game with hidden action, villagers in this study are asked about their reciprocal behavior toward seven potential opponents with different levels of trusting belief. Guilt sensitivity is elicited from the threshold belief to switch from selfish to reciprocal behavior. It appears that males exhibit higher guilt sensitivity. I also find robust supporting evidence for guilt aversion but not for pure altruism or trustworthiness; guilt-averse villagers can borrow from and repay to community members after a disaster. Individuals also suffer less from property crime in villages with a higher guilt-sensitivity neighborhood. However, guilt sensitivity is uncorrelated with contribution to community events. A potential reason for the insignificant effect is discussed.