A Hybrid Discrete Choice Model of Differentiated Product Demand with an Application to Personal Computers

In this paper I construct a new discrete choice model of differentiated product demand that combines the pure characteristics demand model and the random coefficient logit demand model. In this hybrid model consumers are exposed to an idiosyncratic taste shock when they make a brand choice but free from it when they make a product choice within a brand. The simplest version of this model describes an oligopoly market where firms sell vertically differentiated products. Simulations show that the hybrid model is closer to the random coefficient logit model with respect to markup, while it is closer to the pure characteristics model with respect to welfare changes. These three models are applied to data on personal computers to estimate consumer demand. The hybrid model is better identified than the other two models such that coefficients are statistically significant and the willingness to pay for quality improvement is more reasonable. Markup and consumer welfare are also compared to evaluate implications of each model.