Carsharing stated preference (SP)

Abstract

A recent study aimed to estimate the potential of carpooling in Switzerland. Part of this study was a survey in which the attitude of the public towards this transport option was investigated using both multi-response questions and stated preference (SP) experiments. In order to gain an insight on how innovative modes are perceived in general, the SP part was composed of two different experiments, one of them including carsharing as alternative. In the first experiment respondents were choosing among car, public transport, carpooling as driver and carpooling as passenger. In the second experiment respondents were choosing among car, public transport and carsharing. This paper reports on the multinomial logit choice models, which were estimated based on participants’ responses. Both SP experiments were based on a trip reported by participants during a phone interview. For each experiment two specifications, a linear and a nonlinear one were estimated. The nonlinear specification allows investigating the impact of selected socio-demographic variables, in this case income and travel time, on the parameters of the models and on willingness to pay indicators. Such indicators permit to complement the qualitative discussion of the results with quantitative analyses and provide a useful background for policy evaluation and planning. Experiment 2: Carsharing In the second SP experiment the alternatives considered were car, public transport and carsharing. In this case all respondents received the same set of alternatives while the number of situations was limited to six. An issue which had to be tackled was the cost of carsharing travel. The norm in SP experiments with car and public transport as available modal options is to take into account the cost of the ticket for public transport travel and the cost of the gasoline for car travel. It is commonly accepted that gasoline cost is what car users perceive as the cost of a trip with that mode. The parking cost can be eventually added. In the case of carpooling this was appropriate, since carpooling implies the use of a private car. In the case of carsharing the usage fee covers other costs which are not usually taken into account in such experiments, nor generally by the driver a of a private car as cost of a particular trip; car insurance and amortization costs are the most important. For that reason, in the second SP experiment total kilometer costs were used. The kilometer cost was calculated using appropriate tables available on the web page of a Swiss automobile club (14). In order to have personalized costs, twelve different categories were considered according to the type of car (using price as proxy, with four levels) and to the yearly mileage (with three levels). Consumption, as in the previous exercise, was the one declared by the respondent. The cost for carsharing usage was calculated using the current prices of the Swiss operator Mobility (15). The carsharing car was, as far as possible, of the same or similar category as the respondent’s own car. Mobility’s fleet includes most, but not all car types. Another issue is how to take into account the duration related part of the carsharing fee. Carsharing users, in general, pay a fee, which is the sum of a distance dependent fee and a duration dependent fee. The latter depends on the rental time, which broadly corresponds to the duration of the round-trip tour; at least in the case of carsharing systems like Mobility which do not allow one-way rentals. Ideally, one would compare tours and not trips; however, since it was not possible to have the precise information needed for the whole tour, the experiment is based on a choice at the trip level. The ranges for the second experiment are reported in Table 2. They were chosen with the same criteria as for the first experiment.