An integration of hybrid choice and random regret-based models

Abstract

Increasingly, the number of private car tends to exacerbate environmental problems such as emissions, decreasing exhaustible resources, taking land for road and parking spaces, and so forth. The severity of these damaging effects is even more problematic when it is realized that private cars spend the majority of time parked. Public transit could be a good alternative, though it has several shortcomings in terms of door-to-door accessibility, flexibility, personalization and carrying goods. In this context, carsharing systems have recently attracted considerable attention in many cities. These systems provide people the benefits of private cars without the costs and responsibilities of car-ownership. The shared-cars and their operating systems are owned and maintained by carsharing organizations. In order to use a shared-car, people need to join a carsharing organization with deposit and/or monthly membership costs. After that, they can use a shared-car by checking the availability of shared-cars, booking one of them, getting and returning the car at the same designated point-of-departure and paying the costs, including time-based, distance-based and fuel costs. According to the previous research, carsharing systems induce people to sell their own-vehicles and avoid purchasing a new vehicle. Therefore, knowledge of underlying decision-making processes related to joining carsharing system can help designing more attractive systems, thereby alleviating the environmental problems caused by private cars. Several studies have applied state choice experiments to investigate the intention of joining carsharing systems. Although these studies have produced some interesting findings, they have in common some critical limitations. First, prior studies did not consider any alternative behaviors to joining carsharing programs. One of the alternative behaviors is purchasing a new car. If the purchase and operation costs of a new car are reasonable compared to the membership and operation costs and the burdensome procedures of using a shared-car, people might prefer purchasing a new car to joining carsharing. The other alternative behavior is using the current transportation modes: own-car and/or public transit. People may not be interested in joining a carsharing program if current transportation systems sufficiently satisfy their needs. Second, uncertainties underlying carsharing decisions have been largely ignored in previous studies. Since a car is shared by a number of members, there is some inherent uncertainty in obtaining a car whenever they want. In addition, it is more difficult for people to anticipate the expected travel costs by a shared-car than those of a private car or public transit because the usage cost of a shared-car depends on various types of unit cost, which are greatly different from a private car and public transit, and are more sensitive to activity-travel durations. Therefore, it might be expected that these uncertainties play important roles in decisions to join carsharing. The present study develops an integrated model by incorporating random regret into a hybrid choice modeling framework in order to simultaneously consider satisfactions with the current transportation system and the uncertainties in the carsharing decision. Random regret is employed to account for people’s decision-making behavior under the uncertainties of the carsharing system. The model postulates that people choose an alternative which minimizes anticipated random regret. By integrating these modelling approaches, the proposed model allows us to simultaneously understand the effects of the satisfaction of the current system on the carsharing decision under the uncertainties. The parameters are estimated based on a stated choice data about the intention of joining carsharing against purchasing a new car and using the current systems. Before completing the stated choice experiment, respondents were asked the qualities of their current travel modes and their satisfaction with the systems. Some of these actual qualities were shown in the stated choice experiment as the attributes of the current systems and utilized as reference points to set attributes levels of a new car and a shared-car. The knowledge obtained by our empirical study would be useful to establish more effective strategies to increase the number of users of carsharing programs.