City CarShare: Assessment of Intermediate-Term Travel-Behavior Impacts

Author(s): Cervero, Robert; Creedman, Nina; Pai, Madhav; Pohan, Muhammad; Tsai, Yu-Hsin | Abstract: Some nine months after the introduction of City CarShare, a car-sharing program in the city of San Francisco, an estimated 7 percent of trips made by the program's members involved City CarShare vehicles, up from around 2 percent just six months earlier. At the nine-month mark, more than 20 percent of members' vehicle miles traveled (VMT) represented car-share vehicles, a substantial jump from what it was earlier.Evidence suggests that access to car-share vehicles is stimulating motorized travel. Most members do not own cars and many appear to be leasing vehicles in lieu of walking and biking. Car-share vehicles are used more for personal business and social-recreational travel than non-discretionary, routine travel such as to work or school. Cars generally are not used frequently during peak periods or to dense settings well-served by transit, like downtown. In this sense, car-sharing appears to be stimulating a resourceful form of "automobility." Users are accruing substantial travel-time savings and willingly pay market prices for these benefits.Survey results also suggest that car-sharing is cutting into private car usage, especially among higher income members. This appears to be less because members are getting rid of cars and more due to them selectively substituting City CarShare vehicles for their own. Predictive models revealed that the likelihood of car-share usage increased with members' personal incomes, educational levels, and age. Also, members were more likely to lease vehicles if they lived in zero-car households.For the intermediate-term analysis, findings were generally more interpretable than from the short-term analysis. This was expected in that three to four months into the program, when the short-term surveys were compiled, San Francisco's City CarShare was still in its infancy. By the ninth month of the program, many members likely had settled into a certain pattern of usage. This bodes favorably for the ability to firmly gauge impacts one to two years into the program.