Limited information and advertising in the US personal computer industry

Traditional discrete-choice models assume buyers are aware of all products for sale. In markets where products change rapidly, the full information assumption is untenable. I present a discrete-choice model of limited consumer information, where advertising influences the set of products from which consumers choose to purchase. I apply the model to the U.S. personal computer market where top firms spend over $2 billion annually on advertising. I find estimated markups of 19% over production costs, where top firms advertise more than average and earn higher than average markups. High markups are explained to a large extent by informational asymmetries across consumers, where full information models predict markups of one-fourth the magnitude. I find that estimated product demand curves are biased toward being too elastic under traditional models. I show how to use data on media exposure to improve estimated price elasticities in the absence of micro ad data.