Multinomial Probit with Time-Series Data: Unifying State Dependence and Serial Correlation Models

Abstract

This paper develops a general method for treating discrete data sets containing individuals that have made more than one choice under varying stimuli. The multinomial probit model is shown to possess properties that make it very attractive for this application, as with it, it is possible to develop an estimation process that uses all the information in the data, and is both relatively inexpensive and consistent with utility maximization. The method, which is a generalization of Heckman's binary model, can include taste variations and more than two alternatives.