|
Abstract:
|
We propose a new multinomial spatial probit model that connects different consumers using multiple weighted networks, each based on unique similarity structures, that, in turn, influence the correlations between different subsets of the parameter vector. We propose a novel parameter estimation method based on Monte-Carlo Expectation-Maximization (MCEM) which, relative to the standard model, significantly increases number of consumers and choice alternatives that can be handled. We establish the convergence properties of the proposed MCEM algorithm, present computational perspectives on its scalability and provide a distributed computing-based implementation that yields parameter estimates and their standard errors. We apply the model to sales data for compact cars from the Sacramento market and show that incorporating networks based on geographic location and fuel efficiency of previously owned cars deliver more accurate estimates of a consumer's hybrid purchase probability and provide useful insights into the competitive position of the different vehicles in this market. Finally, we demonstrate how a manager can use Conquest Cash strategy to accelerate adoption of Toyota hybrid.
|
