The economics of automated public transport: Effects on operator cost, travel time, fare and subsidy

Highlights

• Study the effect of automation for optimal vehicle size, frequency, fare, subsidy and economies of scale.

• Model applied to electric vehicles with data from Chile and Germany.

• Automation benefits operators and users.

• The benefits of automation are larger in developed countries.

Abstract

It is currently unknown in which city environments, automated vehicles could be deployed at reasonable speeds, given safety concerns. We analytically and numerically assess the impact of automation for optimal vehicle size, service frequency, fare, subsidy and degree of economies of scale, by developing a model that is applied for electric vehicles, with data from Chile and Germany, taken as illustrative examples of developed and developing countries. Automation scenarios include cases with partial driving cost savings and reduced running speed for automated vehicles. We find that a potential reduction in vehicle operating cost due to automation benefits operators, through a reduction of operator costs, and also benefits public transport users, through a reduction on waiting times and on the optimal fare per trip. The optimal subsidy per trip is also reduced. The benefits of vehicle automation are greater in countries where drivers’ salaries are larger.

Introduction

Almost 40 years ago, J.O. Jansson showed that crew costs were by far the largest cost item of local bus companies in Sweden, accounting for 42% of the total operator cost, followed by bus capital costs, which represented 21% of the total costs (Jansson, 1980). The large role of driver wages within the cost structure of urban bus transport does not seem to have changed much over the years. Depending on bus type, driver cost accounts for between 40 and 70 per cent of total bus operator cost in Singapore (Ongel et al., 2019) and Australia (own calculation based on ATC, 2006). In Japan, driver salaries account for 53% and 70% of total operating costs of buses and taxis, respectively (Abe, 2019). In developing countries, where wages are relatively lower, driver cost is less significant but still sizeable, e.g., around 1/3 of total bus operator cost in Santiago de Chile (Librium, 2013). Therefore, it is expected that vehicle automation, where deployment is possible, could have profound impacts on the public transport industry and service in the next decades.

In this context, automated vehicles have the potential to eliminate one of the main elements that cause economies of scale in public transport: drivers’ wages. The cost advantage of placing many travellers in large vehicles, such as buses or trams, will be reduced; thus, shared mobility services with smaller vehicles are expected to play a larger role in a future of highly or fully automated vehicles. Some empirical estimations of the dramatic effects of automation on reducing the costs of motorised shared mobility have been made. For example, in Zurich automation is estimated to reduce the cost of taxi trips by 85% (Bösch et al., 2018) and in Singapore, total operator costs of an electric 6-m long shuttle bus are reduced by 70% if automated, as compared to its human-driven equivalent (Ongel et al., 2019). More conservative estimations are provided by Wadud (2017) for the United Kingdom, who, after assuming that with automation a 40% of current driver costs will still be needed, estimates cost savings of 30% for the taxi industry and between 15% and 23% for the truck industry.¹

Given the large initial cost of the technology to provide full automation capabilities to vehicles, automation is expected to be firstly adopted in ride-hailing, shared mobility and commercial services, rather than for individual ownership and use (Wadud, 2017, Sterling, 2018). This issue has several key implications for the future of mobility, as current research shows that the energy consumption and environmental effects of the future deployment of automated vehicles crucially depend on whether the use of automated vehicles will be individual or mostly shared (Wadud et al., 2016).

Pilot programmes with small scale public transport services have been operating in the past 3–4 years in the form of automated shuttle buses in several countries, such as Switzerland, France, The Netherlands, Sweden and Finland (for a review, see Ainsalu et al., 2018). The first full size 12-m automated buses are scheduled to start trials in 2019 in Singapore² and 2020 in Sweden³ and Scotland,⁴ in all cases being electric vehicles. Parallel to the progress of pilots with automated vehicles for shared use, there is a current debate among experts and researchers on whether fully automated vehicles will ever operate at acceptable levels in urban environments (Kyriakidis et al., 2019). Presently, there is a larger consensus that highly automated vehicles will be able to operate under certain conditions, such as segregated roads and low-speed environments (Kyriakidis et al., 2019).

Beyond operating cost savings, automation is expected to impact public transport in various ways. Several automation technologies will be introduced in public transport services, such as collision avoidance, lane-keeping, bus platooning, bus precision docking (i.e., having a narrow and stable gap between the vehicle and the platform at bus stops), cooperative adaptive cruise control (CACC) and automated emergency braking (Lazarus et al., 2018, Lutin, 2018). The expected benefits of such innovations include a reduction of collisions, injuries and liability costs, improved services for people with reduced mobility and an increase in transport capacity, especially in dedicated infrastructure, such as bus lanes and segregated corridors (Lazarus et al., 2018). Lutin (2018) predicts a great reduction in cost and improvement in service in the US paratransit industry, which caters for persons with reduced mobility, due to automation, even though a fully automated service for mobility-impaired passengers poses further challenges as, e.g., robotic assistance will be required for boarding and alighting.

Three works closely related to the current paper are Abe, 2019, Zhang et al., 2019 and Fielbaum (2019). In Abe (2019), the total cost savings due to automation are estimated for the taxi, bus and rail industry in Japan, including operator cost and travel time for users. The author assumes a waiting time, which is exogenously set and is used for both human-driven and automated vehicles; therefore, the effect of automation on optimal supply levels for a public transport service (e.g., service frequency) is not considered. On the other hand, Zhang et al. (2019) optimise a fleet of fully-automated and semi-automated buses on a hub-and-branch network. By semi-automated buses, the authors consider vehicles forming connected platoons in which only the leading bus has a driver. Bus frequency and size are optimised to minimise total operator plus user cost. Results show that automated services have a larger optimal bus frequency and a smaller vehicle size if the cost savings due to automation compensate for any reduction in speed from automated vehicles. Cost savings of semi-automated vehicles are less prominent than those of fully automated vehicles. Zhang et al. (2019) assume that the linear relationship between bus size and operator cost found by Jansson (1980) still holds with automated vehicles and their numerical application uses data from diesel buses from the previous decade (2000s) in Australia. Finally, Fielbaum (2019) optimise a network composed of a trunk system and feeder lines, in which different configurations of truck systems are compared, following Fielbaum et al. (2016). The author adapts the cost data of automated vehicles estimated in Bösch et al. (2018) to the case of Santiago and finds that automated vehicles provide more “direct” lines, with fewer transfers than human-driven public transport, due to the saving in operator costs. In Fielbaum (2019), only the case of full driving cost saving due to automation and the same running time of automated and human-driven vehicles is considered.

We see that the current understanding of the economics of automated public transport is limited in a number of ways. None of the previous authors analyses the effects of vehicle automation on optimal pricing and subsidy decisions of public transport provision. In this paper, the effect of automation on public mobility services is addressed with a supply optimisation model that takes into account both user and operator costs (Mohring, 1972). Thus, we extend earlier cost models for automated vehicles that focus on operator costs only (e.g., Stephens et al., 2016, Bösch et al., 2018, Ongel et al., 2019), with the inclusion of users’ costs in the form of waiting and in-vehicle times. We go beyond the works of Fielbaum (2019) and Zhang et al. (2019) by analysing the effects of vehicle automation, not only on optimal vehicle size and service frequency but also on optimal fare and subsidy. The degree of scale economies with and without automation is also calculated. Unlike Zhang et al. (2019), we use updated data from the operation of electric vehicles, given that all current public transport pilots of automated vehicles utilise electric vehicles and this is the technology expected to prevail (over internal combustion motor vehicles) at least in the near future.

Unlike Fielbaum (2019) and Zhang et al. (2019), we make our own estimation of capital and operating costs of automated and human-driven vehicles from scratch (see Appendix), which proves to be relevant as we are able to numerically assess if assumptions made by previous authors, concerning the effect of automation on the marginal cost of increasing vehicle size, hold. We analyse alternative scenarios of deployment of automated vehicles, considering the cases in which not all driving costs are saved due to automation, and that running speed of automated vehicles might be lower than that of human-driven vehicles, due to safety concerns in cities (Zhang et al., 2019 also analyse this case). Furthermore, this is the first article to compare the effect of automation on the optimal design of a public transport service in developed and developing countries – for which Germany and Chile are chosen for illustration – specifically concerning differences in drivers' salaries and values of time. We are able to show that, given a reasonable set of assumptions on operating and capital costs of human-driven and automated electric vehicles, Jansson's linear relationship between vehicle capacity and cost (Jansson, 1980) holds with human-driven and automated electric vehicles, for a range of vehicle types from cars to articulated buses.

We focus on fixed-route services. The service frequency (i.e., the inverse of the headway between vehicles) and the vehicle size are optimised to minimise total costs. Therefore, the choice of vehicle types, such that standard car, van, minibus and standard bus, is endogenous in the model, which is solved for increasing levels of demand. We also determine the effects of automation on the optimal (first best) fare, subsidy and on the degree of economies of scale of public transport provision. The increased capital cost of vehicles due to automation will be accounted for together with the reduction in operating cost due to automation. Sensitivity analyses on key parameters are performed to understand the main determinants of optimal shared or public transport supply levels.

In terms of results, the contributions of this paper are the following. It is shown that automation will increase the demand threshold that justifies the adoption of bigger vehicles, and that the size of the effect of automation on reducing vehicle size and increasing frequency depends on the country where automation is applied and on the final conditions regarding cost savings and running speed with automated vehicles. Second, combinations of running speed and cost saving with automation, that determine if there is an effect of automation on optimal frequency and vehicle size, are numerically established, which serve as a frontier for observable automation effects on supply outputs. Third, the relative cost saving due to automation is larger in Germany than in Chile, but in both countries, large savings are expected if full automation eventually materialises. Scenarios in which not all driving cost are saved and running speeds of automated vehicles are low can significantly reduce expected cost gains. Fourth, we show analytically and numerically that the reduction in vehicle operating cost due to automation benefits two parties: (i) operators, through a reduction of operator costs and (ii) public transport users, through a reduction on waiting times and on the fare to be paid for the service. Moreover, there is a reduction in the optimal subsidy per trip to be allocated to the public transport system. The size of these savings in some cases is straightforwardly estimated and in others depends on the parameters of the problem. Automation reduces the degree of economies of scale in public transport. Numerically, we find that for automation to have a noticeable effect on reducing optimal fares, a fraction larger than 50% of the current driving cost must be saved.

The rest of the paper is organised as follows. Section 2 summarises current research on cost effects of vehicle automation and the use of electric vehicles for public transport. In Section 3, the supply optimisation model is presented, together with the derivation of optimal price and subsidy rules, which are used as a base to theoretically analyse the effect of automation on optimal supply and pricing outputs. Section 4 presents the estimation of relevant cost and operation parameters for the cities of Munich in Germany and Santiago in Chile, and the effect of automation on operator cost parameters is assessed. The full optimisation model is solved and applied in Section 5. Section 6 concludes.