Title: Reputation Dynamics in a Market for Illicit Drugs

Abstract: We analyze reputation dynamics in an online market for illicit drugs using a novel dataset of prices and ratings. The market is a black market, and so contracts cannot be enforced. We study the role that reputation plays in alleviating adverse selection in this market. We document the following stylized facts: (i) There is a positive relationship between the price and the rating of a seller. This effect is increasing in the number of reviews left for a seller. A mature highly-rated seller charges a 20% higher price than a mature low-rated seller. (ii) Sellers with more reviews charge higher prices regardless of rating. (iii) Low-rated sellers are more likely to exit the market and make fewer sales. We show that these stylized facts are explained by a dynamic model of adverse selection, ratings, and exit, in which buyers form rational inferences about the quality of a seller jointly from his rating and number of sales. Sellers who receive low ratings initially charge the same price as highly-rated sellers since early reviews are less informative about quality. Bad sellers exit rather than face lower prices in the future. We provide conditions under which our model admits a unique equilibrium. We estimate the model, and use the result to compute the returns to reputation in the market. We find that the market would have collapsed due to adverse selection in the absence of a rating system.