Longitudinal Analysis of Car Ownership and Car Travel Demand in Paris Metropolitan Area Using a Pseudo-panel Data Approach
Benoît CORNUT, IFSTTAR-DEST
The aim is to find the determinants of car ownership and use and determine income and fuel price elasticities of demand by comparing different areas of residence and income groups. Pseudo-panel data and a model with a semi-log specification are used.
Car ownership and use have reached a maximum in Paris region. Car ownership stagnates at above one car per household and car mobility has reached a maximum at 1.5 trips per person and per day which implies a decrease of the modal share of car trips from 44% in 2001 to 38% in 2010. However, this automobile ceiling has not appeared at the same time for the inhabitants of different zones, more or less distant from the city center. Indeed, we note since 1990 a demotorisation and a reduction of the average number of trips made by the inhabitants of the city of Paris and the same phenomenon has appeared in the inner suburb in the 2000s. Finally, the rise in motorisation and use continues only in the outer suburbs.
This phenomenon raises the question of the saturation of demand and notably the potential decorrelation of car ownership and use from economic variables (income, fuel price). That’s why we will interest in this paper to the analysis of longitudinal behaviour for car ownership and car travel demand at household level. More precisely, the research is seen from an inequality point of view. We will determine income and price elasticities of demand for different residential densities of location and income classes. Thus, the objective is to exhibit contrasted reactions of different categories of household over time. Moreover, the longitudinal analysis will allow us to determine life-cycle effects and generation effects.
In order to analyse longitudinal behaviour, panel data should be used. However, there is a lack of panel data in many countries and these data also present a problem of attrition which limits long term analysis. That is why the use of pseudo panel data approach can be a good alternative and is employed in this paper. Indeed, the pseudo-panel approach is a relatively new econometric method first introduced by A. Deaton in 1985 which allows longitudinal analysis without genuine panel data. The principle is to group individuals or households into cohorts in using repeated cross-sectional data. Cohorts are built from time-invariant variables such as year of birth, gender or level of education. Attention has to be paid to the size of each cohort and also to its homogeneity. To create cohorts, sufficiently large number of observations in each cohort (generally at least one hundred observations is suggested) are necessary to avoid a biased estimation. Although year of birth is the most often used variable to create cohorts, other complementary variables can be used to make the cohorts more homogenous. The aim is to reduce the variation within the cohorts and increase the variation between the cohorts while having a number a observations large enough.
Our study case is Paris metropolitan area which includes territories with contrasted densities and a differentiated access to public transport. Pseudo-panel data are constructed in using a succession of five large independent surveys (Enquête Globale Transport) conducted between 1976 and 2010 where more than 15.000 households have responded to each survey. In our case, the cohorts of households are grouped by years of birth at 3 years interval.
Concerning the modelling, we have estimated two models (for car ownership and car travel demand) having a semi-log linear specification. The dependant variables used are the number of cars per household and the number of trips made by car per household and per day. Exogenous variables introduced are the logarithm of constant income per household, real fuel price taking into account the evolution of petrol-diesel repartition in the car fleet over time, the life-cycle effects with the age of the head of household in 6 age bands, family composition (number of adults and children) and the residential location (Paris, inner and outer suburbs). In the pseudo-panel approach, the common procedure is to calculate the mean of each variable at cohort level. However, precautions have to be taken for the model estimation in order to have convergent estimators.
To estimate the model, cohort-specific intercept terms have been included in the equation so it can be seen as a least-square dummy variables model (fixed-effects model) and be estimated by the OLS procedure. This method permits to have a consistent estimation. In our case, the cohort-specific intercept terms can be interpreted as generation effects.
Association for European Transport