ABSTRACT
After the outbreak of COVID-19 pandemic, crowding has been highlighted as a risk factor for contracting acute respiratory infections (ARIs) such as COVID-19, which has affected the demand for public transport. Although several countries, including the Netherlands, have implemented differential fare systems for peak and off-peak travel to reduce crowding during the rush hours, the problem of overcrowding on trains has remained prevalent and is expected to cause more disutility than even before the pandemic. A stated choice experiment in the Netherlands is conducted to understand the extent to which people can be motivated to change their departure time to avoid crowded trains during rush hours by offering them real-time information on on-board crowding levels and a discount on the train fare. To gain further insights into how travelers respond to crowding and capture unobserved heterogeneity in the data, latent class models have been estimated. Unlike the previous studies, the respondents were segregated into two groups before the start of the choice experiment based on their indicated preference to schedule a delay earlier or later than their desired departure. To study the change in travel behavior during the pandemic, the context of different vaccination stages was also provided in the choice experiment. Background information collected in the experiment was broadly categorized as socio-demographic, travel and work-related factors, and attitudes towards health and COVID-19. It was found that the coefficients obtained for the main attributes which were presented in the choice experiment (on-board crowd levels, scheduled delay and discount offered on full fare) were found statistically significant, and in line with previous research. It was concluded that when most of the people are vaccinated in the Netherlands, the travelers become less averse to on-board crowding. The research also indicates that certain groups of respondents, such as those who are highly crowd averse, and are not students, can be motivated to change their departure time if real-time crowding information was provided. Other groups of respondents who were found to value fare discounts can also be motivated to change their departure by similar incentives.
ABSTRACT
The performance of ride-sourcing services such as Uber and Lyft is determined by the collective choices of individual drivers who are not only chauffeurs but private fleet providers. In such a context, ride-sourcing drivers are free to decide whether to accept or decline ride requests assigned by the ride-hailing platform. Drivers’ ride acceptance behaviour can significantly influence system performance in terms of riders’ waiting time (associated with the level of service), drivers’ occupation rate and idle time (related to drivers’ income), and platform revenue and reputation. Hence, it is of great importance to identify the underlying determinants of the ride acceptance behaviour of drivers. To this end, we collected a unique dataset from ride-sourcing drivers working in the United States and the Netherlands through a cross-sectional stated preference experiment designed based upon disparate information conveyed to the respondents. Using a choice modelling approach, we estimated the effects of various existing and hypothetical attributes influencing the ride acceptance choice. Employment status, experience level with the platform, and working shift are found to be the key individual-specific determinants. Part-time and beginning drivers who work on midweek days (Monday-Thursday) have a higher tendency to accept ride offers. Results also reveal that pickup time, which is the travel time between the driver’s location and the rider’s waiting spot, has a negative impact on ride acceptance. Moreover, the findings suggest that a guaranteed tip (i.e., the minimum amount of tip that is indicated upfront by the prospective rider, a feature that is currently not available) and an additional income due to surge pricing are valued noticeably higher than trip fare. The provided insights can be used to develop customised matching and pricing strategies to improve system efficiency. Since the study has been conducted during the COVID-19 crisis, the potential implications of the pandemic on ride acceptance behaviour have been examined using an Integrated Choice and Latent Variable (ICLV) model. The results show that drivers with a higher sensitivity to the COVID-19 effects tend to have a lower acceptance rate.