A Dynamic View of Heterogeneity in Car Travel and of Its Main Determinants – An Analysis from the French Car Fleet Survey



A Dynamic View of Heterogeneity in Car Travel and of Its Main Determinants – An Analysis from the French Car Fleet Survey

Authors

Richard GRIMAL, French Technical Department Of Studies On Transport, Roads And Bridges

Description

This paper is focused on improving our understanding of the diffusion process of car travel in France, the analysis being based on panel data from the French Car Fleet Survey.

Indeed, accounts for travel behaviors and travel demand models are often based on aggregate time-series or cross-sectional views, thus failing to capture either time-related or individual heterogeneity of car travel. Through the combination of dynamic econometric modeling and variance analysis, we will study the influence of the changing distribution of financial, spatial and driving constraints (license holding, individual motorization...) on the organization of individual heterogeneity in car travel, the hierarchy of its determinants, and the distribution between observed and unobserved heterogeneity.

Cross-sectional and dynamic income- and price-elasticities of car travel will be estimated for different time periods. These results will be related to the historical development of car travel, leading to suggest alternative specifications where car travel is gradually slowing down from a certain level, or is even reaching saturation on the long-run.

We will pay specific attention to the 2000's, as there are characterized by a stagnation of car travel after several decades of uninterrupted growth, mainly because of harder economic conditions bringing a reversal in the long-term trend towards ever-releasing constraints. We will try to determine whether the stagnation of car travel is a standard reaction to the shift in external conditions, or is the sign of some behavioral transition, towards a declining income sensitivity, a greater resort to alternative modes and car travel rationalization.

247 words

Abstract

This paper is focused on understanding the diffusion process of car use in France, the analysis being based on panel data from the French Car Fleet Survey, covering the period from 1984 to 2010.

Parameter estimates associated with explanatory variables in travel demand models are often based either on aggregate time-series or cross-sectional views of individual heterogeneity. The first type of data allows to capture the effect of aggregate factors varying over time such as average fuel prices and incomes, driving license holding rates or land use patterns, while the second is rather providing evidence for inter-individual heterogeneity in travel behaviours at a given period of time, in relation with exogenous constraints such as lifecycle position, residential location or financial resources. However, none of these data simultaneously allows to capture temporal and individual heterogeneity, leading to possibly biased and time-contingent parameter estimates. For instance, evidence was found from National Transport Surveys that travel behaviours were decreasingly heterogeneous with respect to lifecycle position or income, and increasingly heterogeneous with respect to residential location, between low- and high-density areas. Furthermore, time-series differ substantially from cross-sectional estimates of income elasticities, also declining over time with increasing incomes and car use levels already attained.

Through the combination of dynamic econometric modelling and variance analysis, we will bring elements of understanding for the temporal variability of income, residential location, and lifecycle effects. To be more specific, we will study how the changing distribution of financial, spatial and driving constraints (driving license holding, motorization), is affecting the level and organization of individual heterogeneity in car travel, either through the hierarchy of its determinants, or through the distribution between observed and unobserved heterogeneity, respectively illustrating the main well-known constraints shaping travel behaviors, and permanent unobserved characteristics such as preferences, skills, etc.

By the way, we will display cross-sectional and dynamic estimates of income- and price-elasticities for car travel. The temporal variability of elasticities will be interpreted by putting the development of car travel back into its historical framework: starting from a simple model where elasticities are supposed to be constant, estimates will then be provided for alternative specifications. For instance, we can justify formulations of demand for car travel into which it should gradually slow down for various reasons (substitution of car by higher-speed modes, decreasing utility of additional travel), or even converge towards a saturation threshold as external constraints are gradually releasing (relatively stable daily trip frequencies and time budgets altogether with a limited substitutive potential of car to the other modes). Under this framework, we are expecting the influence of releasing or strengthening constraints on cross-sectional estimates to depend both on the stage of the diffusion process (from starting up to levelling off), and on its extension (general and homogeneous, or heterogeneous between social groups/residential locations).

Eventually, we will pay specific attention to the 2000’s, as this period is characterized by a breaking trend in car travel, levelling off after several decades of uninterrupted growth. For a large amount, it can be related to a reversal in the dynamic of constraints, with harder economic conditions (increasing fuel prices and stagnating incomes) following a long period where constraints in terms of residential location, travel costs or access to driving had been releasing. By using previously estimated econometric models, we will try to assess whether the stagnation of car travel can be depicted as the result of stable behaviors, i.e a standard reaction to a dramatic shift in external conditions, or is rather reflecting some transition in the behavioral framework, taking the shape of a lower sensitivity of car travel to improved economic conditions, a greater resort to cheaper modes such as transit, walking and cycling, and car travel rationalization either through the renouncement to some unnecessary trips or through closer destinations.

The results of this research can nourish the national debate and reserach about the future of transport demand. They should also contribute to improving the design of transport models, by simultaneously accounting for cross-sectional and time-related heterogeneity, the distributional aspects of constraints and their interdependencies. Eventually, by identifying permanencies and discontinuities in behaviors, it will provide a better understanding of the historical breakdown in car travel.

691 words

Publisher

Association for European Transport