Some Methods for the Testing and Estimation of Dynamic Models Which Use Panel Data

Publisher Website
Google Scholar
Abstract
In this paper the authors address some of the inferential problems posed by longitudinal data on the discrete choice behaviour of a collection of individuals. In particular, an integrated framework is developed which enables heterogeneity and nonstationarity to be explicitly included in stochastic models of binary choice behaviour. The emphasis is upon minimum assumptions about the nature and determinants of heterogeneity and nonstationarity, as any uncontrolled variations may result in the identification of spurious adaptive behaviour. Moreover, all the models are readily calibrated and tested using widely available computer software. Some assessment is made of the flexibility of the modelling framework with respect to its potential for handling attrition in panel membership, more extensive choice sets, and exogeneous variables.