Traditional models of consumer choice assume consumers are aware of all products for sale.This assumption is questionable, especially when applied to markets characterized by a high degree of change, such as the personal computer (PC) industry. I present an empirical discrete-choice model of limited information on the part of consumers, where advertising influences the set of products from which consumers choose to purchase. Multi-product firms choose prices and advertising in each medium to maximize their profits. I apply the model to the US PC market, in which advertising expenditures are over $2 billion annually. The estimation technique incorporates macro and micro data from three sources. Estimated median industry markups are 19% over production costs. The high industry markups are explained in part by the fact that consumers know only some of the products for sale.Indeed estimates from traditional consumer choice models predict median markups of one fourth this magnitude. I find that product-specific demand curves are biased towards being too elastic under traditional models of consumer choice. The estimates suggest that PC firms use advertising media to target high-income households, that there are returns to scope in group advertising, and that word-of-mouth or experience plays a role in informing consumers. The top firms engage in higher than average advertising and earn higher than average markups.
