Abstract

We investigate simple online behavior patterns that suggest the current dualistic view of simultaneous vs. sequential search modes is likely founded on coarse data. In contrast with the predictions of these search modes, we find consumers are selective about the attributes they inspect, they revisit items to acquire additional information, and often convert without collecting all data available on the selected alternative.

Obtaining the correct demand structure is pertinent because the search literature has documented that consumer search modes can largely determine welfare allocations. We consider the full dynamic problem faced by consumers searching for used vehicles at an online car dealership and propose a feasible simulation-based estimator to characterize consumers' behaviors. We use the recovered fundamentals to analyze the consequences of different information provision activities – where a seller selects which attributes to feature and which to relegate. Sellers can do better than maximizing consumers' information by featuring characteristics that tend to pool “bad news” with ex-ante attractive alternatives, taking advantage of consumers unwillingness to search all data available. We find information provision tends to have modest impact on welfare, with the preferences of seller and consumers over information layouts often coinciding.