Estimation of Heterogeneous Preferences: An Empirical Analysis of Demand for Internet Services

This paper presents a structural econometric framework for discrete and continuous consumer choices in which unobserved preference heterogeneity is modeled explicitly. The model is applied to analyze data from the U.C. Berkeley Internet Demand Experiment (INDEX). This experiment records participants’ ISDN line usage for Internet access under various price-quality combinations. A structural utility model for users is estimated. The model allows for unobserved intra-personal and interpersonal preference heterogeneity. Users’ ex ante valuations of the service, before they use it, can be identified separately from users’ online service valuations. The model performs well in out-of-sample predictions and can be used to simulate responses to alternative pricing scenarios.