Advertising and dynamic demand for differentiated products

Date of Completion

January 2008


Business Administration, Marketing|Economics, Agricultural




This dissertation consists of three essays on topics in empirical industrial organization. The first essay proposes and estimates a fully dynamic model of generic and brand advertising in a differentiated product oligopoly. It first specifies a nested logit demand system incorporating brand and generic advertising goodwill stocks. Demand side parameters are then used to calibrate a dynamic game of brand and generic advertising that takes into account the vertical relationship between manufacturers and retailers. Estimates from the fluid milk product category indicate that brand advertising is effective for increasing brand level demand and generic advertising has a differential effect on individual brands. On the supply side, we found that it is optimal for brand manufacturers not to advertise in the presence of generic advertising. ^ The second essay proposes a semiparametric Bayesian framework for the analysis of discrete choice models that can be applied to both individual and aggregated data. Instead of assuming a parametric model for the distribution of consumer sensitivity to product characteristics as is typically the case in both Bayesian and classical models, the distribution of consumers' heterogeneity is modeled using a Dirichlet process prior which varies with consumer demographic characteristics. A Markov chain Monte Carlo algorithm is developed for fitting the model, and the methodology is illustrated using a panel dataset of peanut butter purchases. ^ The third essay adapts the model developed in Essay 2 to a structural random coefficients discrete choice model with learning, state dependence, and long lasting advertising effects. It applies the model to analyze demand for ready-to-eat breakfast cereal in selected U.S. cities. The model parameter estimates are then used to investigate brand-level dynamic advertising competition. Results indicate that consumers have negative priors about new cereal brands but after experiencing with the new products, they found these to have better quality than the established brands. The estimated distribution of individual-specific parameters uncovers some of the limitations of the commonly used parametric models of consumer heterogeneity. ^