The Impact of Fake Reviews on Reputation Systems and Efficiency

Online interactions are frequently governed by reputation systems that allow users to evaluate each other after an interaction. Effective reputation systems can increase trust and may improve efficiency in market settings. In recent years, however, fake reviews have become increasingly prevalent. Since it is difficult to clearly identify fake reviews in field studies, we design a laboratory experiment. Using a repeated public good game with a reputation system, we study (i) how feedback manipulation influences the reliability of average ratings and (ii) whether the existence of manipulated ratings reduces efficiency. We find that feedback manipulation generally decreases the reliability of average ratings in comparison to a control treatment where cheating is not possible. When manipulation is possible and free, average ratings become less reliable, expectations are lower and both cooperation and efficiency are significantly reduced. When there are costs of manipulation, however, average ratings are more reliable and contributions and efficiency are not impaired. Interestingly, this is the case even when costs of manipulation are comparatively low.