Speaker: Professor John Geweke (Distinguished Professor, Centre for the Study of Choice, University of Technology Sydney)

Topic: Nonparametric Bayesian Modeling of Monotone Preferences for Discrete Choice Experiments

Abstract:
Discrete choice experiments (DCE’s) are widely used to about the distribution of consumer preferences across product attributes. Such experiments are often designed and conducted deliberately for the purpose of designing new products or predicting post-merger market structure. In web-based purchase environments DCE’s can be embedded in the alternatives offered to consumers. There is a long-standing literature on nonparameteric and Bayesian modeling of preferences for the study of consumer behavior when there is a market for each product, but this work does not apply when such markets fail to exist as is the case with most product attributes. This paper takes up the case in which attributes can be quantified and preferences over these attributes are monotone: for example, credit cards, for which key attributes include the interest rate on outstanding balances, annual fees, credit limits, length of window with no interest payment, and degree of acceptability. It shows that a monotone nondecreasing (but not necessarily quasiconcave) utility function is appropriate in these circumstances. The paper introduces a Dirichlet process prior for marginal utility and demonstrates that all monotone nondecreasing utility functions are supported by the prior. MCMC simulation is used to access the prior distribution, and the paper demonstrates how the posterior sample can be integrated into a decision support system. The paper will include an example either with simulated or actual data.

For more information about upcoming seminars, please click here.