A dynamic approach to car availability throughout the life course

Abstract

The relatively new research field of mobility biographies designates the analysis of long-term mobility behaviour and the availability of mobility tools in a life span. A retrospective survey of the TU Dortmund, ETH Zurich and Goethe University Frankfurt collects data on individual mobility biographies of three different generations in a household with a life-calendar. Most of the past long-term decisions made by individuals such as buying a house or changing job affect their preferences in future periods and induce economic constraints in the form of transaction costs. Ignoring these aspects may lead to biased estimates in the analysis. A dynamic probit model is used to identify impacts on the individual decisions on car availability in a life span and tests for differences between gender or is used to include the time dependency of the explanatory variables such as age, the number of children or education. The focus of the paper is to compare the modelling results following common practices in the life course calendar literature, based on random effects probit models with the results obtained with a dynamic random effects probit model with autocorrelation. In contrary to the classic random effects probit model approach the main advantage of the dynamic probit approach is to explicitly model the correlated time-fixed and time-varying unobserved heterogeneity by the means of composite marginal likelihood estimation.