Unintended consequences of corruption indices: an experimental approach

Using the results of a large-scale (N=900) online experiment, this paper investigates how the information about a group corruption level may harm intergroup relations. Corruption indices are widely used as a measure of quality of governance. But in addition to be a valuable tool for investors and policy makers for making informed decisions, they may also result in statistical discrimination: people from a more ‘corrupt’ region may be perceived as less trustworthy or more inclined to dishonest behavior. We manipulated the amount of information people have about three different Russian regions in two simple behavioral games (‘Cheating game’ and Trust game). In a Cheating game after the main stage where they report whether they observed a head or a tail on a flipped coin, they guessed how many participants in each of the three regions reported more personally profitable outcome (heads) as well as make trasfer decisions in a standard trust game. In the baseline treatment we provided them with a set of generic information about each region (such as population size), while in the main treatment they were additionally informed about the degree of perceived corruption in each region. The presence of corruption information made people substantially overestimate the degree of dishonesty in more ‘corrupt’ regions and decreased the trust towards a person from this region. The results demonstrate the potentially harmful unintended consequences of corruption indices that have to be taken into account by policy makers.