Kitamura, Ryuichi (1991) A Review of Dynamic Vehicle Holdings Models and a Proposal for a Vehicle Transactions Model. Institute of Transportation Studies, University of California, Davis, Presentation Series UCD-ITS-RP-92-01
Household vehicle holdings observed at a time point embody many decisions made by the household over a period of time. The set of vehicles the household holds, or "household vehicle fleet," is not acquired instantaneously; it is the result of a series of transaction decisions to acquire, replace, or dispose of household vehicles. Each of these decisions is conditioned on the current vehicle holdings, and reflects the long-term planning effort of the household.
Most household vehicle holdings models are either "cross-sectional," using data obtained at one point in time, or "pseudo-dynamic," using repeated cross-sectional data or aggregate time-series observations obtained from various sources. Both classes of models are subject to certain limitations. Application of cross-sectional models to forecasting implies "longitudinal extrapolation of cross-sectional variations" which is valid only under very restrictive conditions. For example, questions have been raised whether cross-sectional elasticities estimated from these models are identical to longitudinal elasticities that are associated with changes in behavior. Cross-sectional models' ability to accurately represent household behavior under changing income, fuel prices, traffic congestion, etc., should be critically re-examined. Aggregate time-series-models, on the other hand, are often incapable of capturing causal relationships governing the behavior of individual behavioral units, thus tend to be limited in their accuracy and policy sensitivity.