Investigating the determinants of a Peer-to-peer (P2P) car sharing. The case of Milan Ilaria Mariotti Paolo Beria Antonio Laurino DAStU, Politecnico di Milano SIET 2013 Venezia, September 18th – 20th , 2013 STRUCTURE • Aim • Literature review on P2P • Data and methodology • Descriptive statistics • Econometric analysis • Discussion and conclusions AIM • Investigate the main determinants to join a P2P car sharing system by means a descriptive statistics and two discrete choice models: binomial logit model and multinomial logit model 1,129 Milan citizens have been surveyed (Green Move project). Literature review (1) • Ex-post analyses on Car Sharing (CS) prevail • Main determinants to join CS: ▫ density of users aged 25 – 45, single or living in small households ▫ well educated with an income higher than the average ▫ cost sensitive ▫ environmentally conscious ▫ good public transport service ▫ CS mainly used for recreation/social activities Literature review (2) • Literature on P2P system is scanty ▫ Hampshire and Gaites (2011) emphasise the higher accessibility that P2P scheme could entail, in particular in lower density areas, thanks to the almost total absence of the upfront costs that a traditional CS operator has to bear to buy its fleet. ▫ Hampshire and Sinha (2011) analyze the main trade-off of balancing car utilization with reservation availability. Data and methodology • Dataset – Green Move survey conducted in 2012 among the inhabitants of the municipality of Milan (1,129 respondents) • The probability to undertake a P2P carsharing is investigated by means of a descriptive statistics, which results are corroborated by a binomial logit model and a multinomial logit model Dependent variable Answers Yes, with all people that joined the service Yes, but only with an entourage of people I choose Yes, but only with my neighbours Yes, but only with my colleagues No, because the car is a personal effect No, because I want the car always available No, because I do not need to deprive me of my car Answers – Multinomial logit Yes, with all people joining the service Answers – binomial logit 1 Yes, with the people I know (friends, neighbors and colleagues) 2 No 0 Yes 1 No 0 * question: “Would you be interested, under these conditions (…) to share your car (or one of your cars) at the time of the day you indicate?” Explanatory variables Socio economic Description Gender Dummy variable: 1 “ if male, 0 “ if female. Age Age of the respondent Education Dummy variable: 1 “ if the respondent achieved a bachelor degree, “0 otherwise Number of cars owned by the family Green Attitude Travel behaviour Variable District of residence N. of owned cars Oil price Modal choice: - LPT, Bike, Foot, Motorcycle, Car (driver), Car (passenger) Daily travel by car for: - reaching the workplace,or the LPT stop - moving within the neighbourhood or outside - leisure in the city, other motives Car sharing member Area C tool and travel behaviour change Dummy variable: 1“ if the respondent has changed his/her travel patterns, 0“ otherwise. District where the respondent lives. Dummy variable. Six dummy variables suggesting the main modal choice adopted by the respondent. Six dummy variables underlying why the respondent uses the car daily or very often. Dummy variable: 1“ if the respondent is or has been member of CS services, 0 “ otherwise. Dummy variable: 1 “ if the respondents have reduced the car use consequently the Area C introduction, 0“ otherwise Descriptive statistics (1) • 53.4% potential sharers 4% 6% 35% 55% All P2P members Group of people Neighbours Colleagues Descriptive statistics (3) Respondents’ travel behavior LPT Bike Foot Motorcycle Car-driver Car-passenger Potential sharers 26.6 11.2 15.5 6.7 35.9 4.0 Non- sharers 24.1 6.4 15.4 5.1 42.7 6.2 9% of the potential sharers are or have been members of the Milan CS vs. 2.5% of the non users Binomial logit model Model 1 Model 2 Model 3 -0.0124*** -0.0121** -0.0123** Gender 0.2174* 0.2158 0.1980 Degree 0.2701*** 0.2705** 0.2502* Number of owned cars LPT 0.2794*** 0.3652*** 0.2853*** 0.2915* 0.2856*** 0.3217* Bike Foot 0.6610*** 0.1597 0.6638*** 0.1688 0.6579*** 0.1663 Motorcycle 0.3271 0.3107 0.3104 Car (driver) -0.0058 -0.0067 0.000 Car (passenger) Carsharing Member -0.1482 0.9872*** -0.1637 0.9772*** -0.0949 0.9994*** Area C- car use reduction Oil price increase -car use reduction To reach the workplace 0.3317*** 0.3397*** 0.3473*** 0.5079*** 0.5066*** 0.5306*** -0.0998 -0.1132 -0.8179*** 0.4661** 0.0927 -0.0729 -0.8079*** 0.4410* 0.1050 -0.0677 -0.7691** 1129 1129 1129 -730.3661 -727.9935 -722.9772 0.0636 0.0666 0.0730 Age LTP stop Neighbourhood Leisure in the city Constant n. obs. Log Likelihood PseudoR2 Results Group 1 GROUP 0: Those not interested to join a P2P CS system Model 1 Model 2 Model 3 -0.001 -0.000 -0.0010 Gender Degree 0.568*** 0.428*** 0.581*** 0.437*** 0.5601*** 0.3936*** Number of owned cars 0.374*** 0.377*** 0.3850*** LPT 0.609*** 0.516*** 0.5282*** Bike 0.931*** 0.942*** 0.9268*** Foot Motorcycle 0.003 0.499 0.021 0.489 0.0072 0.4720 Car (driver) 0.214 0.226 0.2449 0.302 0.950*** 0.305 0.931*** 0.3823 0.9593*** 0.207 0.212 0.2189 0.403*** 0.406*** -0.205 0.4362*** -0.2114 0.562** 0.265 0.5230* 0.2747 -0.043 -2.9049*** -0.0262 -2.8665*** Group 1: all members Age Car (passenger) CS Member Area C- car use reduction Oil price increase -car use reduction To reach the workplace LTP stop Neighbourhood Leisure in the city Constant -2.8898*** Results Group 2 GROUP 0: Those not interested to join a P2P CS system Model 1 Model 2 Model 3 -0.0186*** -0.0184*** -0.0185*** Gender 0.0255 0.0191 0.0039 Degree 0.1834 0.1817 0.1679 Number of owned cars LPT 0.2192*** 0.2264 0.2263*** 0.1652 0.2227*** 0.2022 Bike 0.5014*** 0.4990*** 0.4949*** Foot 0.2253 0.2293 0.2309 Motorcycle Car (driver) 0.2241 -0.1246 0.2024 -0.1337 0.2035 -0.1359 Car (passenger) -0.4143 -0.4354 -0.3788 CS Member Area C- car use reduction 0.9938*** 0.3979*** 0.9871*** 0.4055*** 1.0102*** 0.4147*** Oil price increase -car use reduction 0.5673*** 0.5669*** 0.5903*** To reach the job place -0.0391 -0.0559 LTP stop Neighbourhood 0.3984 -0.0053 0.3836 -0.0104 Leisure in the city Constant -07010 -0.0819 -0.6882 -0.0834 -0.6434 1129 1129 1129 -1107.8923 -1104.2871 -1096.0491 0.0548 0.0579 0.0649 Group 2: Friends, neighbours Age n. obs. Log Likelihood PseudoR2 Results (1) The probability to join a P2P CS is positively and significantly related to: ▫ ▫ ▫ ▫ ▫ users’ education (bachelor degree), car ownership (more than two cars), travel behaviour (LPT and bike), CS membership (previous or present), cost sensitiveness (i.e. oil price increase). Results (2) When comparing the users willing to share their own car with all members of the P2P system (confident shares), it results that they tend to be: ▫ ▫ ▫ ▫ male, use the car daily to reach the LPT stop, have reduced the car use because of the Area C, are less willing to live in zone 9. While, those willing to share their own car only with a selected group of people, tend to be: ▫ younger, ▫ use the bike to travel, ▫ are less willing to live in zone 7. CONCLUSIONS • Relevance of the three groups of determinants: socioeconomic, travel behavior and green attitude. • Potential users are sensitive to CS systems – being or having being members of the Milan CS –, and are costsensitive (i.e. oil price increase and Area C policy tool). Besides, they prefer to ride the bike or use the LPT to travel. Questions and suggestions are welcome Ilaria Mariotti DAStU – Politecnico di Milano ilaria.mariotti@polimi.it