Collaboration Models for Ride-hailing and Transit Service Providers to Facilitate First- and Last-mile Connectivity

Abstract

Mobility-as-a-Service (MaaS) has gained significant popularity in the past few years. Various ride-hailing service providers (RSPs, e.g., Uber, Lyft, Didi, Ola, etc.) have entered the transportation market to provide this service. They offer door-to-door connectivity, online booking and payment facility, customizable rides, and other such features to the travelers. All these attractive features have caused a substantial increase in the mode shares of the RSPs. However, this rising popularity has raised a concern among the city transportation planners. Quite often, the RSPs operate their vehicles at an occupancy which is much less than the full capacity. As a result, more vehicles are required to serve the traveler demand which has the potential to cause traffic congestion on roads. This leads to wastage of non-renewable fuel as well as increased travel times and carbon dioxide emissions due to the vehicles. On the other hand, public transit is a potentially cheaper and sustainable mode option. However, in many cases, transit stops are located far from the travelers' residential and/or activity locations. This hinders the extensive usage of public transit as travelers may have to walk for long distances or arrange an intermediate mode of transport to/from transit stops. Therefore, the travelers shift to other modes, such as personal vehicle, ride-hailing service, etc., which are more comfortable options.

Due to these issues, there is an interest in exploring the collaboration between the RSPs and the transit agencies. In such a collaborative mobility service, the RSPs will provide connectivity between the travelers' residential/activity locations and the transit stops (first- and last-mile connectivity) and the transit agency will facilitate travel on the long-haul part of the journey. Such a collaborative service can potentially bring together the best of both operators' services by providing seamless door-to-door travel while keeping the overall travel prices low and reducing traffic congestion. However, in many cities, barring a few exceptions, neither the RSPs nor the transit agencies appear eager to co-operate with each other. This is because the scientific literature lacks evidence and methods to formulate, evaluate, and facilitate discussions for such a collaboration. Therefore, the objectives of this thesis are to formulate models for evaluating collaboration between the RSPs and the transit agencies in the context of first- and last-mile services. In particular, we focus on optimal pricing models to assist the RSPs and the transit agencies in pricing the collaborative mobility service to enhance the mode shares and profits of both operators.

The thesis is divided into three parts. In the first part, we propose a game theory and discrete choice theory based tri-level collaboration model. The proposed model is a Stackelberg game with the bus agency as the leader and the RSPs as followers. It is applicable to a stylized four-node network with one origin-destination (O-D) pair. All the players in the proposed game have the objective of profit maximization as a function of their price in the collaborative service. Needless to say, the optimal price setting should consider the travel mode choice preferences of the travelers. We apply the proposed model to two major travel corridors of Bengaluru, a major metropolitan city in India. The simulation results indicate promising results in the form of increased profits and mode shares of the collaborating agencies and improved travel times due to a reduction in congestion. The travelers also benefit in the form of a better expected maximum value of utility in the context of mode choice. The new collaborative mode turns out to be a well-balanced alternative in terms of travel price and travel time (lower travel time than the walk and bus mode combination and lower travel cost than the direct RSP mode between travel origin and destination).

In the second part of the thesis, we extend the model formulated in the first part for one O-D pair to a city-scale network with multiple O-D pairs. This model is also a tri-level model, formulated using the concepts from game theory, random utility maximization based discrete choice theory, and traffic network equilibrium theory. It is applicable to a network with one bus route. In this model, the bus agency and the RSPs play a Stackelberg game to set the price for their part in the collaborative service. We apply the city-scale model to the Sioux Falls test network to test the robustness of the model. Also, we apply the model to a sub-network in the Bengaluru city. The simulation results for both the networks corroborate the findings in the previous part of the thesis. The RSPs and the transit agencies observe a higher profit and effective person trips due to collaboration. There is also a slight improvement in the average speeds and a small reduction in the carbon dioxide emissions due to the vehicles.

In the third part of the thesis, we further extend the city-scale model to include an auction component. This helps the bus agency to optimally select RSPs for collaboration. In this model, the bus agency conducts an auction inviting price bids from the RSPs which are defined as the amount per traveler corresponding to the collaborative service that the RSPs are willing to pay/receive from the bus agency for collaborating. Based on the submitted bids, the bus agency announces the winners of the auction. Later, the winning RSPs and the bus agency play a Stackelberg game to set prices for their part in the collaborative service. We apply the model to the Sioux Falls test network and compare the optimal pricing model with- and without-auction. For the setting under consideration, we do not observe a significant difference between the two cases. However, the auction is able to increase the competition on the supply-side (between RSPs) which can potentially result in better services for the travelers.