Finding seller quality and quality sellers in online markets: an agent-based analysis

We construct an agent-based model of the online electronics market, consisting of 10,000 buyers and 50 sellers. Buyers repeatedly interact with sellers, searching for high quality and low price, but only some buyers know seller quality prior to purchase. Buyers may learn seller quality directly or from information obtained from other buyers or else estimate quality through market share heuristics. We examine the efficacy of these learning and estimation methods and find that preferential searching, where the probability of searching any seller is its market share, leads to higher consumer surplus and significantly lower levels of price dispersion.