Abstract
We develop a model of search by imperfectly informed consumers with unit demand. The innovation is that consumers learn spatially: sampling the payoff to one product causes them to update their payoffs about all products that are nearby in some attribute space. Search is costly, and so consumers face a trade-off between "exploring'' far apart regions of the attribute space and ``exploiting'' the areas they already know they like. Learning gives rise to path dependence, as each new search decision depends on past experiences through the updating process. We present evidence of these phenomena in data on online camera purchases, showing that the search paths and eventual purchase decisions depend substantially on whether the past items searched were surprisingly good or bad. We argue that search intermediaries can affect purchase decisions not only by highly ranking products that they would like purchased, but also by highlighting bad products in regions of the attribute space that they would like to push the consumer away from.