Car Ownership in Great Britain - a Panel Data Analysis
HANLY M and DARGAY J, ESRC Transport Studies Unit, University College London, UK
In a paper presented at the Transportation Research Board Conference in Washington D.C. in January of this year, the authors developed a dynamic panel car ownership model for households in Great Britain using data from the Bdtish Household Panel Survey (B
In a paper presented at the Transportation Research Board Conference in Washington D.C. in January of this year, the authors developed a dynamic panel car ownership model for households in Great Britain using data from the Bdtish Household Panel Survey (BHPS). In the introduction to that paper, the authors re-iterated the much vaunted advantages of panel data for analysis i.e. that panel data can capture changes at a level of detail that is necessary to capture true causality. They stated that this is useful for the detection of, among other intertemporal effects, response lags and experience effects. The present paper extends the analysis of the eadier model by examining whether past experience has a real effect on current expedence (of owning a car), or whether observed experience effects are due to unobserved differences between households.
Dependence on past experience can be incorporated into a dynamic model by introducing lagged endogenous variables. However, it is important to determine if this apparent dependence of currant choices on past choices is "genuine" state dependence or whether it is in fact due to unobserved differences or heterogeneity between households. If behavioural differences are mainly due to unobserved factors across households, which are otherwise observationally equivalent, and if these unobserved factors are invadant over time but correlated with the measured explanatory vadables, the estimated model coefficients will be biased unless the heterogeneity is not taken into account. In this paper a relatively simple type of heterogeneity is modelled by modifying the error term of the estimated equation to include a household-specific time invadant component following the model in Kitamura and Bunch (1990).
Association for European Transport