Corruption Bias and Information : A Study in the Lab

Our study examines whether actual corruption, measured by individuals direct experience of corruption episodes (bribery), matches their perceptions of the phenomenon. Our experimental participants play a repeated public good game with mandatory minimum contribution and are given the possibility to bribe a computerized bureaucrat in order to free-ride. We elicit beliefs about the perceived level of corruptibility of the bureaucrat and others' corruption attempts. We study participants' willingness to corrupt and the gap between perceived and actual corruption under two information conditions. Results show that, although anonymous, spreading news about an attempt of corruption is enough to discourage such attempts, lowering the corruption rate. Consequently, when receiving no information, participants expect others to corrupt more, raising the index of perceived corruption