Predicting Take-up of Servicised Automobility (car Clubs and Cars-on-demand) Using a Strategic-tactical Discrete Choice Model with Composite Alternatives



Predicting Take-up of Servicised Automobility (car Clubs and Cars-on-demand) Using a Strategic-tactical Discrete Choice Model with Composite Alternatives

Authors

S Le Vine, A Sivakumar, J Polak, Imperial College London, UK; M Lee-Gosselin, Universite Laval, CA and Imperial College London, UK

Description

Servicised automobility (encompassing car clubs and cars-on-demand) is a small but rapidly-growing form of personal travel, principally though not exclusively found in Western societies.

Abstract

Servicised automobility (encompassing car clubs and cars-on-demand) is a small but rapidly-growing form of personal travel, principally though not exclusively found in Western societies. Predicting the trajectory of this consumer market is challenging in a number of respects, some of which are typical of the introduction of new products whilst others are unique to this context.

This paper begins by discussing the proposed approach to address two methodological challenges. First, given the relative cost structures of using a servicised car or a personal car, for any given journey it is nearly-always easier and less expensive to drive a personal car. The question is rather how competitive a lifestyle incorporating servicised automobility is in comparison to one of car ownership; traditional mode-choice analysis ignoring this level of choice-making is inadequate. The second challenge is that people seem to use servicised cars for a rather small portion of their overall mobility pattern, as a ?gap-filler? in which other non-car modes are used for the majority of journeys. A discrete choice model framework is presented which addresses this choice context by specifying that people choose a portfolio of mobility resources (owning a car, owning a public transport season ticket, or subscribing to a car service, for instance), and that this choice is a function of a set of travel needs rather than a single journey. In other words, the choice set is specified to be amongst ?composite? alternatives, each of which is a different combination of multiple ?elemental? alternatives.

The paper then illustrates several unique statistical properties of the proposed model. A subset of alternative-specific parameters which are in principle identifiable in traditional mode-choice models are shown to not necessarily be identifiable in the proposed specification. The IID assumption is shown to be quite problematic in this type of model structure.

Results from the empirical application are then presented, which consist of predictions of take-up and usage of two forms of servicised automobility in Greater London. It then concludes with, in addition to a summary of this research, a brief discussion of issues raised for further enquiry.

Publisher

Association for European Transport