Dynamic Modeling of an Individual's Discrete Choice Behavior on Continuous Time Frame Using Panel Data: Toward Dynamic Demand Analysis without the Assumption of Equilibrium*

Abstract

A dynamic model of an individual's commute mode switching behavior along a continuous time dimension is presented. The model is specified as a semi-Markov process model of commute mode use. Its parameters are estimated using discretetime panel data. Parameter estimates indicate that the duration of carpooling to commute is shorter than those of the other commuting modes. The average speed of each commute mode has effects on commute-mode switching behavior. A sample enumeration study using this dynamic behavioral model offers estimates of the dynamic characteristics of aggregate commute-mode shares, and the time it will take for the market to reach an “equilibrium.” The numerical study thus reveals that the market is not in equilibrium even for commute travel modes.