Towards a dynamic discrete choice model of fleet size and vehicle type

Vehicle ownership at the level of the individual household is an important topic that has attracted research attention since the mid seventies. To date studies have concentrated on either choice of fleet size or choice of vehicle type. In all instances the models have been static. The approach to vehicle ownership adopted in the thesis departs from past reported practices in two important respects; both elements of the overall vehicle ownership decision are modelled (fleet size and vehicle type) and the analysis is inter temporal. A linked system of structured logit models is developed to explain vehicle holdings at a particular point in time and adjustments in the holdings over time. The main objective in applying the models to a 5 year retrospective data panel of Sydney households is to test a number of important hypotheses. In particular we seek to (i) establish the underlying structural relationship between choice of fleet size and choice of vehicle type in the static choice context, (ii) identify the main influences on vehicle ownership choice behaviour, (iii) assess the importance of experience effects in the adjustment process and (iv) gauge the temporal stability of individual parameter estimates and entire models. The study successfully introduces a temporal dimension to the individual choice modelling framework, provides a linking mechanism for temporally adjacent decisions and in the static choice context overcomes a number of deficiencies shared by past reported efforts to model the same or similar choices (a).