Mobility tools and use: Accessibility s role in Switzerland

Size: px

Start display at page:

Download "Mobility tools and use: Accessibility s role in Switzerland"

Stuart Freeman
5 years ago
Views:

1 Mobility tools and use: Accessibility s role in Switzerland A Loder IVT ETH Brisbane, July 2017

2 In Swiss cities, public transport is competitive if not advantageous. 22 min min min 2

3 And between the largest cities? Text 1h 47min 1h 32min 1h 29min 3

Total annual costs of mobility Fix and variable car costs for 15 000 km/year:

4 Total annual costs of mobility Fix and variable car costs for km/year: CHF/USD Nation-wide season ticket for unlimited travel: CHF/USD

5 Motivation Modeling multi-modal travel behavior in Switzerland Owning a car and a season ticket Owning a car, but using public transport Accounting for similar accessibility levels by both modes (accessibility as a measure of the generalized cost of travel) 5

Data: Transportation Survey 2010 Nation-wide 62 868 persons from 59 771 households Demographic household information and vehicle holdings

6 Data: Transportation Survey 2010 Nation-wide persons from households Demographic household information and vehicle holdings Respondents report on socio-economic status mobility tool ownership one day travel diary of a randomly assigned day transport policy attitudes 6

7 Data: Accessibility I - Concept Gravitation-based approach (Hansen, 1959): Accessibility A i at location i to all opportunities O j at locations j A i = N j=1 O j exp(βt ij ) with t ij travel time from i to j and β decay parameter For Switzerland (Sarlas, Fuhrer, & Axhausen, 2015) β car = β pt = O j for two cases: population and workplaces 7

and public transport lines included 2949 zones in

8 Data: Accessibility II - Travel times Macroscopic traffic model (PTV VISUM) AADT calibrated for 2010 All major roads and public transport lines included 2949 zones in Switzerland (census districts) Road network Public transport network 8

9 Data: Accessibility III Results Four accessibility measures (two modes x two opportunities) High correlation between all measures: Principal component analysis Three comprehensible principal components General accessibility Comparatively better accessibility by public transport Comparatively better accessibility to workplaces 9

10 First principal component 10

11 Second principal component 11

12 Model Joint modeling of heterogeneous outcomes (Bhat, 2015) Binary: Ownership of mobility tools Count: Number of trips; generalized ordered response (Castro, Paleti, & Bhat, 2012) Endogenous effects (car ownership on car use) Estimation with maximum likelihood (MSL/MACML) Programmed in Stata 14 using ml 12

13 Results I: Mobility tool ownership and use Mobility tool Number of trips Car Season Public Nonmotorized Car ticket transport General accessibility *** *** *** *** Better accessibility by public transport *** *** Better accessibility to employment *** *** Local access to public transport Very good (base) Good *** *** Moderate *** *** Low *** *** Very low *** *** Spatial typology City (base) Agglomeration *** *** *** *** *** Countryside *** *** *** *** *** Endogenous effects Car always available *** *** *** Subscription to season ticket *** *** *** Constant *** *** *** *** *** * p < 0.05, ** p < 0.01, *** p < 0.001, socio-demographic control variables not shown. Likelihood of choosing public transport options increases with greater levels of accessibility. 13

14 Results II: Car-sharing Two level season ticket type choice between local and nation-wide season ticket (GA) (Becker, Loder, Schmid, & Axhausen, 2017) Any season Car-sharing GA Car ticket club General accessibility *** *** Better accessibility by public transport ** *** Better accessibility to employment *** *** At least good access to public transport *** *** * Spatial typology City center(s) *** *** base Agglomeration *** Isolated city * Countryside *** * p < 0.05, ** p < 0.01, *** p < 0.001, socio-demographic control variables not shown. Better accessibility by public transport favors the nationwide ticket. 14

15 Outlook Update and comparison with 2015 survey Residential self-selection (Cao, Mokhtarian, & Handy, 2009) Accommodating spatial interactions (Bhat, Pinjari, Dubey, & Hamdi, 2016) Commuters choice of mobility tool ownership: considering residential and workplace location 15

16 Conclusions Use of principal component analysis to highly correlated accessibility measures Modeling multi-modal travel behavior with new accessibility variables: good predictor Simulation of changes in multi-modal travel behavior with changes in one mode s travel times 16

17 Thank you for your attention! 17

18 References I Bhat, C. R. (2015). A new generalized heterogeneous data model (GHDM) to jointly model mixed types of dependent variables. Transportation Research Part B: Methodological, 79, Bhat, C. R., Pinjari, A. R., Dubey, S. K., & Hamdi, A. S. (2016). On accommodating spatial interactions in a Generalized Heterogeneous Data Model (GHDM) of mixed types of dependent variables. Transportation Research Part B: Methodological, 94, Becker, H., Loder, A., Schmid, B., & Axhausen, K. W. (2017). Modeling car-sharing membership as a mobility tool: A multivariate Probit approach with latent variables. Travel Behaviour and Society, 8, Cao, X. J., Mokhtarian, P. L., & Handy, S. L. (2009). Examining the impacts of residential self selection on travel behaviour: A focus on empirical findings. Transport Reviews, 29,

19 References II Castro, M., Paleti, R., & Bhat, C. R. (2012). A latent variable representation of count data models to accommodate spatial and temporal dependence: Application to predicting crash frequency at intersections. Transportation Research Part B: Methodological, 46(1), Hansen, W. G. (1959). How Accessibility Shapes Land Use. Journal of the American Institute of Planners, 25, Sarlas, G., Fuhrer, R., & Axhausen, K. W. (2015). Quantifying the agglomeration effects of Swiss public transport between 2000 and paper presented at the 15h Swiss Transport Research Conference, Ascona, April

20 Backup 20

21 Correlations among unobserved common factors (M1) r21 Car season ticket *** r31 Car car trips ** r41 Car public transport trips *** r51 Car non-motorized trips ** r32 Season ticket car trips *** r42 Season ticket public transport trips *** r52 Season ticket non-motorized trips r43 Car trips public transport trips *** r53 Car trips non-motorized trips *** r54 Public transport trips non-motorized trips * p < 0.05, ** p < 0.01, *** p < Both modes are substitutes in ownership and use. Correlations between ownership and use within each mode indicates commitment. 21

22 Correlations among unobserved common factors (M2) r21 Any season ticket car *** r31 Any season ticket car-sharing club ** r41 Any season ticket GA *** r32 Car char-sharing club *** r42 Car GA *** r43 Car-sharing club - GA * * p < 0.05, ** p < 0.01, *** p < Both modes are substitutes. Sample selection (any season ticket/ga) cannot be neglected. GA and car-sharing club membership are complements. 22

Preferred citation style

Preferred citation style Axhausen, K.W. (2016) How many cars are too many? A first attempt, presentation at 16 th COTA International Conference of Transportation Professionals, Shanghai, July 2016.. How