Last Minute Feedback

Feedback mechanisms in electronic markets allow trading partners to rate each other after a transaction and to observe past ratings given by other market participants. Such mechanisms are considered crucial for the success of internet trading platforms, where alternative governance methods are often missing or too expensive. However, the concern has been raised that on eBay positive feedbacks may be given in order to encourage reciprocation, and negative feedbacks might not be given because of the fear of retaliative negative feedback. If quantitatively important, these practices substantially reduce the informational content of reputation indicators. In this paper, we investigate the micro structure of the eBay feedback mechanism in light of an analysis of the sequencing and timing of feedbacks. Since the transactions per se are unobservable, we derive bounds on probabilities of buyer and seller satisfaction by exploiting alternative sets of identifying assumptions. We argue that substantive bias is potentially accumulated in the reputation measures and conclude with easy-to-implement suggestions geared at improving the design of the feedback mechanism in terms of the informational content of its records. JEL Classification: D44, L15, L86.