AN EMPIRICAL-INVESTIGATION OF THE CONSISTENCY OF NESTED LOGIT-MODELS WITH UTILITY MAXIMIZATION

Global conditions under which nested logit models are consistent with utility maximization are provided by Daly and Zachary and by McFadden. Recently, Borsch-Supan and Herriges and Kling have provided conditions under which a nested logit model is locally consistent. However, previous work has not discussed implementation of these conditions. Here, three alternative approaches to checking and formally testing for the consistency conditions using classical statistics are investigated. In addition, a Bayesian approach to interpreting and imposing the local consistency conditions is provided. The application is based on anglers' choices regarding sportfishing alternatives in southern California.