THE TIMING OF CHANGE: DISCRETE AND CONTINUOUS TIME PANELS IN TRANSPORTATION

Abstract

Individuals adapt to changed circumstances at various points in time. Panel data typically are collected at discrete points in time. How well can models constructed from discrete time panel data approximate the underlying process or timing of change? This is an important research issues as we increase our understanding of the benefits of longitudinal data in general and panel data in particular. In this paper we discuss change within a framework of the timing of change and a continuous time metric. The conditions under which a discrete-time approximation is an acceptable representation of the processes which occur in practice in continuous time is considered. Given that panel data are typically collected in discrete time units, the ability to 'translate' discrete time observations into a metric set of continuous time estimates is of interest in the study of the duration of events. We illustrate the suggested approach in the context of a duration model of the timing of switching to a new urban tolled road.