International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Modeling Swiss mobility tool ownership: A hybrid multivariate probit approach with sample selection
Henrik Becker, Allister Loder, Basil Christian Schmid, Kay W Axhausen

Last modified: 28 March 2017



Ownership of mobility tools such as cars, season tickets or car-sharing memberships is a strong determinant of a person’s short-term travel behaviour. But despite their important role, the interrelations between the three mobility tools are not yet fully understood. Whilst the backgrounds of car-ownership and season-ticket ownership and their interdependencies have already been subject to various quantitative studies, the motives for and impacts of car-sharing membership are less clear: Although the literature consistently suggests, that car-sharing reduces its members’ level of car-ownership and use, only few papers have addressed these questions using quantitative methods (e.g. Mishra, 2015).

In Switzerland, residents can hold a subscription to local season tickets, a nation-wide season ticket, called the Generalabonnement (GA), become member of a nation-wide station-based car sharing scheme and own a car. In this choice set, some combinations are more complements (car-sharing and local season ticket), others may be more substitutes (car and GA), whereas still others are mutually exclusive (GA and local season ticket). Given such complex interdependencies, traditional discrete choice models like the multinomial logit or nested logit may not be able to fully capture all of those aspects. Therefore, this research proposes a modeling framework that allows for a choice set of complementary and exclusive choices. The model is based on the multivariate Probit model and incorporates a Heckman-type sample selection (Jenkins et al., 2006; van der Straeten et al., 2003).

Furthermore, past research has shown, that attitudes play a major role in explaining mobility tool ownership (e.g. Beck et al., 2013). In order to include the effect of attitudes, an integrated choice and latent variable approach (Ben-Akiva et al., 2002) was applied by simultaneously modeling the choice of mobility tools and two latent variables based on responses to mobility pricing schemes, policies and infrastructure improvements. The survey captures 1) “car-loving” attitudes exhibiting a general reluctance towards any kind of pricing schemes leaving car users worse off (LV1) and 2) attitudes reflecting a desired increase in public transportation infrastructure investments funded by car pricing revenues (LV2), which themselves are a function of various socio-demographic indicators.



The model is applied to data from the 2005 and 2010 Swiss transportation micro-census in which an 8% subsample of observations responded to attitudinal questions besides their socio-demographics and mobility tool ownership. The Swiss transportation micro-census is a large-scale survey carried out every five years on roughly 1% of the nation’s population.



The framework proposed in this paper is one of the most comprehensive approaches to model mobility tool ownership including both innovative forms of mobility and latent attitudes.

The results from the structural model indicate that younger respondents with low income and education living in households without kids exhibit the highest LV1 values, whilst younger male respondents with lower income, high education and living in larger urban households without kids (e.g. in a student co-op) show the highest LV2 values.

Moreover, the model shows the effect of various spatial, socio-demographic and attitudinal characteristics on the choice of mobility tools. In particular, preliminary results suggest, that the effect of LV1 on the respondents’ mobility tool ownership is much stronger than the effect of LV2. In addition, strong anti-correlations between motorization and season-ticket ownership as well as between private cars and car-sharing are observed. The conditional probabilities and elasticities are presented, allowing further insights on how the three different mobility tools substitute each other.



Beck, M. J., J. M. Rose and D. A. Hensher (2013) Environmental attitudes and emissions charging: An example of policy implications for vehicle choice, Transportation Research Part A, 50, 171–182.

Ben-Akiva, M., D. McFadden, K. Train, J. Walker, C. Bhat, M. Bierlaire, D. Bolduc, A. Börsch-Supan, D. Brownstone, D. S. Bunch et al. (2002) Hybrid choice models: Progress and challenges, Marketing Letters, 13 (3) 163–175.

Jenkins, S. P., L. Cappellari, P. Lynn, A. Jäckle and E. Sala (2006) Patterns of consent: evidence from a general household survey, Journal of the Royal Statistical Society: Series A (Statistics in Society), 169(4), 701-722.

Mishra, G. S., R. R. Clewlow, P. L. Mokhtarian and K. F. Widaman (2015) The effect of carsharing on vehicle holdings and travel behavior: A propensity score and causal mediation analysis of the San Francisco Bay Area, Research in Transportation Economics, 52 (1) 46-55.

van der Straeten, K., A. Trannoy, N. Picard and C. Hagneré (2003) L’importance des incitations financières dans l’obtention d’un emploi est-elle surestimée?, Économie & prévision, 160 (4) 49–78.

Conference registration is required in order to view papers.