An analysis of household vehicle holding durations considering intended holding durations

Household vehicle holding durations are examined in this study using panel data. Panel data enable the observation of changes in household vehicle hol dings in discrete time periods. If the data set contains retrospective reca ll data which offer information on the types and the occurrence time points of the transactions since the last survey, direct observation of the trans action process along a continuous time axis is possible. Furthermore, if th e data contain information on intentions for future transactions and actual outcomes of the intentions in later waves, the relationship between intend ed transactions and actual transactions can be observed. In this study, we develop models of actual vehicle-holding durations and models of intended v ehicle holding durations for the same set of vehicles. Comparing these two sets of models, the effects on household vehicle holding durations of chang es in the household's plans for vehicle holding and unexpected events can b e inferred. To represent unaccounted associations between intended vehicle holding durations and actual vehicle holding durations, vehicle specific er ror components are introduced into the duration models. A non-parametric ap proach is adopted in model estimation using mass points, which requires no assumption on the distribution of the error components. (C) 2000 Elsevier S cience Ltd. All rights reserved.