Bribery, Punishment, Norms and Reciprocal Relationships: Evidence from US and Pakistan

This paper contributes to the corruption literature by implementing bribery in the laboratory as a repeated three person sequential game, consisting of a firm, a government official and a citizen. In contrast to the design of Abbink et al (2002) and Cameron et al (2009a), our design fixes the value of the bribe that the firm can offer and the favor that the government official can grant, allowing us to focus on variable citizen punishment. The experiment consists of two treatments, one with citizen punishment and one without. In both treatments, the firm moves first and decides whether to initiate a bribe, the government official is given the choice of accepting or rejecting, and, in the Punishment treatment, the citizen is able to punish either or both of the two other players. Experimental sessions were conducted in the US and in Pakistan, which have very different levels of corruption. Differences in punishment levels between the firm and the official serve as a measure of the “blame” put on the parties engaged in a corrupt act. We show that in the U.S., the citizens punish the firms slightly less than the government officials across all rounds where successful bribes have occurred, while in Pakistan the firms are punished significantly more than the officials. Overall, firms in Pakistan are punished more than officials, and punishment has virtually no effect on the bribes offered. In the U.S., punishment is shown to lower both the proportion of bribes offered and accepted. Officials in both cultures are sensitive to punishment with U.S. officials slightly less resistant. We conclude that the presence of strong corruption norms along with client patron relationships in Pakistan mean that officials are not seen as accountable for corruption, so that the blame for bribery is attributed to the initiators of bribes.