A reputation for the good stuff: user feedback signaling and the deep web market silk road

Abstract

Despite complete user anonymity, asymmetrical information, and incomplete enforcement mechanisms, the deep web market Silk Road facilitated approximately $200 million in illegal drug sales in 34 months. This study tests how the site's reputation system facilitated successful transactions and how user feedback functioned as the primary signal of seller quality in the absence of formal contract enforcement. Using novel data from the site on marijuana, amphetamine, and meth transactions, listings, and vendors, I find strong evidence that consumers engaged with the site's reputation system and relied on both item and seller-level information. Hedonic regressions provide evidence of a 'bad news' learning environment, estimating a 3 to 11% price discount for negative item reviews. Seller ratings are found to act as an effective proxy for permanent seller characteristics, and named trains act as a primary source of quality (and price) differentiation for marijuana. I find no evidence of price penalties or signal heterogeneity for new sellers. This study is the first to shed light on the value of reputation on the deep web's largest marketplace, yielding new insight into the mechanisms modern markets use to overcome social distance and prevent market failure.