This thesis consists of three empirical essays on topics in industrial organization. Chapter 1 introduces the three essays and describes their main results. In chapter 2, I measure the effects of advertising on households' choices among competing brands of differentiated products when households are habit persistent. In markets characterized by habit persistent consumers, an important role for advertising may be to encourage consumers to "switch" to newer, less familiar brands. This potentially enhances competition between competing brands by reducing household-level differentiation between brands that a household is and is not experienced with, contrary to the traditional Bainian arguments that advertising reinforces market power via product differentiation. Estimates from my household-level brand choice models for the breakfast cereals market imply that advertising not only promotes competition among existing brands by reducing the degree of product differentiation which arises due to experience, but also facilitates entry of new brands. Therefore Bain's argument that advertising is a "product differentiation barrier to entry" seems appropriate only in the presence of production or marketing economies of scope. In chapter 3, which draws upon joint work with Andrea Coscelli, we empirically verify the existence of and quantify the extent of informational barriers to entry into the anti-ulcer drug market. Using an Italian dataset of complete prescription histories for 350 doctors in the Rome metropolitan area, we study the diffusion process of omeprazole, an anti-ulcer molecule which entered the market in 1990. Our estimates ascribe the gradual growth in prescriptions of omeprazole over the sample period is due largely to doctors' initial pessimism about omeprazole's quality. Strikingly, we find that this uncertainty is resolved by first-hand experience (actual prescriptions) rather than through marketing activities and other potential sources of information: this despite including a complete set of dummies for all the months in our sample in order to soak up aggregate effects. In chapter 4, based on joint work with Han Hong, we derive an econometric model of a multi-round ascending (or English) auction. Our work differs from the previous literature in two aspects: (1) we present the first equilibrium econometric model of an ascending auction with both private and common value components that we are aware of; and (2) we allow for bidders' valuations for an object to be non-identically distributed--ours is a model of an asymmetric auction. While the aim of this chapter is mostly methodological, we estimate our model using data from the spectrum auctions run by the U.S. Federal Communications Commission (FCC).
