The Economics of Autonomous Vehicles

Will AVS improve social welfare — or create unprecedented congestion? The jury is still out.

Will AVS improve social welfare or create unprecedented congestion? The jury is still out.

(AVS) are widely expected to radically AUTONOMOUS VEHICLES change mobility patterns and improve the efficiency of our transportation systems. The highest level of automation that is currently being tested allows for vehicles to travel without a human on board. This concept opens up abundant opportunities in the transportation industry, but it also has implications for road capacity — both positive and negative.

As a result of the predicted benefits of AVS — which include smoother traffic and improved safety — more than 50 cities worldwide have committed to deploying them in the near future, and another 27 are preparing for automation by undertaking surveys of regulatory, planning and governance issues raised by these vehicles.

Not surprisingly, the private sector is actively pursuing vehicle automation. By now, most car manufacturers have established an AV division and expect to make the technology available to the mass market as early as 2025. By the year 2045, AV market share is predicted to be as high as 87.2 per cent. In this article we will discuss some of the opportunities and challenges that lie ahead for this burgeoning sector of the automobile industry.

The Current State of the Art

At present, Daimler AG (Mercedes-benz) is considered the leader among the world’s automakers in the development of automated driving technologies. Its Drive Pilot system allows the driver not only to stay within the boundaries of a lane, maintain a safe trailing distance and stop when necessary while in congestion, it also assists with steering, switching lanes and overtaking other cars when the driver signals a lane change. Tesla’s Auto Pilot system, installed on its flagship models, features a similar set of autonomous-driving functions.

Turning to Japan, Nissan Motor Corporation has been the

keenest to develop autonomous technologies, marketing some models installed with an autonomous driving system called ProPILOT, which features keep-in-the-lane and keep-the-distance functions for highway driving and follow-and-stop functions for driving in congestion. Meanwhile, Subaru Corporation has earned recognition for its Eyesight tool — an emergency braking system to help avoid or reduce front-end crash damage as well as keep-in-the-lane and keep-the-distance functions, which are installed on many of its flagship models.

Many car manufacturers have also recognized the emerging ‘shareability’ of AVS and are planning to initiate their own ridesharing programs. For example, Ford recently released a plan to roll out Level-4 AVS designed for commercial ride-sharing applications by 2025; General Motors is also developing automated Chevy Bolts for shared use; and Waymo (which is owned by Google) is partnering with Chrysler to create a shared AV enterprise.

One of the perceived drawbacks of vehicle sharing in the automated age is the increased traffic that will be injected into transportation networks as a result of ‘relocation trips’ — whereby a driverless car returns to its ‘home base’ after completing a ride. In addition to the increased traffic caused by these ‘zombie trips’, regular trips may also increase, as passengers will get to engage in alternative activities while riding. Skeptics of automation claim that AV trips — both regular and zombie — will worsen traffic conditions and clog major urban streets. Some even argue that average vehicle occupancy by humans might get as low as 50 per cent due to zombie trips, and that overall traffic will increase by up to 15 per cent.

In response to the skeptics, advocates of automation believe that AVS will make traffic smoother overall, due to minimized abrupt acceleration and braking, improved communications through vehicle-to-vehicle and vehicle-to-infrastructure channels, and a reduction in accidents. Several studies address this claim. For instance, journalist Clifford Atiyeh has estimated that automation will increase speeds by 23 to 39 per cent under fuel-economy conditions and by eight to 13 per cent in congested traffic. In this view, reductions in accidents will also improve traffic, as 25 per cent of congestion is attributed to traffic incidents.

Moreover, when AVS reach their predicted high marketshare, systemic improvements may appear in terms of intersection controls (i.e. traffic light design). One futuristic vision is to have no traffic lights whatsoever at intersections, but instead, a scenario where cars pass through the area in a synchronized manner without having to stop.

With recent advancements in automation technology, many provincial and state governments in North America, Europe and South Asia are now issuing permits for AVS to drive on designated roadways. Google has tested driverless vehicles over more than two million miles in cities, including Mountain View, Austin and Phoenix. The U.S. is a leader in this testing stage, partly due to its National Highway Traffic Safety Administration, which has issued a set of national guidelines outlining the principles of driverless vehicle pilots. These guidelines streamline the testing phase and motivate companies to pilot their prototypes in reallife traffic conditions. Canada has also begun several pilots in the provinces of Ontario and Quebec by allowing firms such as Uber to run their driverless vehicles on ‘live’ streets.

Broadly speaking, once the testing phase is complete, the next step for AVS will be regulation. There are already 17 states in the U.S. pursuing Av-enabling legislature by passing bills to regulate operations and licensing. Regulatory policies include changing traffic rules to accommodate AVS, land use intervention, starting new ride-sharing services, initiating ‘pilot zones’ (where AVS can be tested), taxi reform (i.e. making taxis automated) and transit automation (e.g. making buses and subways automated). Figure One shows some of the cities that are taking a holistic approach to implementing multifaceted policies. As Indicated Austin, Singapore and Helsinki are among those taking the lead in automation.

Semi-autonomous vehicles already exist in the consumer market, and fully autonomous vehicles are likely to arrive in the next decade.

One approach that is being neglected thus far is subsidization, and we believe this is a mistake. Subsidization — whereby households would be granted rebates for purchasing AVS — is a practical policy approach because most households will not be able to afford AVS due to their high ownership costs and the expensive embedded-autonomy equipment in the vehicle. At the moment, the technology used in AVS includes light detection and ranging systems, sensors, software and other advanced computing power. These components alone can cost more than US$ 30,000 (up to US$ 100,000 for military uses). However, this equipment is expected to become more affordable as AVS become available to the public on a mass scale. Hensley estimates that 15 years after the commercialization of AVS, their cost will drop from a US$ 10,000 markup (i.e. the additional payment for autonomy technology) to a US$ 3,000 markup. Before reaching these affordable markups, however, we believe that governing agencies should promote AV purchases by implementing rebates and subsidies.

The coming trade-off between infrastructure efficiency and induced traffic has caused a heated debate about the benefits of vehicle automation, and experts are taking sides based on speculations — without substantial scientific evidence. Of course, this is partially because AVS are not yet commercialized and no data is not yet available to support either side’s arguments. Nevertheless, the fact is that semi-autonomous vehicles already exist in the consumer market, and fully autonomous vehicles are likely to arrive in the next decade. As indicated earlier, AVS are expected to promote shared mobility because they can be relocated among multiple passengers in driverless mode. By sharing a ‘fleet’ of AVS, each user would pay less for mobility and have a higher incentive to travel due to the extra free time provided. This paradox will potentially lead to more traffic in urban areas.

On the upside, the connectivity features of AVS will enable them to use road space much more efficiently. This efficiency would offset the extra traffic from zombie trips and overall decrease travel times. Going forward, cities should not inadvertently support AVS without an in-depth analysis of their impacts.

Our Research

A number of studies have investigated the impact of AVS in terms of fuel economy, induced traffic, willingness-to-pay, traffic flow, safety, intersection control, parking, and their use as a shared fleet between a group of users. Among the key findings to date, researchers have shown that network capacity generally increases with AV market share because of ‘platoon formation’ and the lower reaction time of AVS. These studies relate capacity to the share of AVS in the network, where the highest capacity is reached when the entirety of traffic is automated.

While research on vehicle automation is growing, the majority of studies to date have taken an engineering approach by looking at either infrastructure management or the travel patterns of AV users in futuristic scenarios. Very few studies have focused on optimal policies for vehicle automation from an operations management perspective. Among those that have, Northwestern Professor Xin Chen et al. propose ‘automated zones’ comprised of a particular set of streets allocated to driverless vehicles. Researchers have also considered policies for managing a fleet of shared AVS and the implications of parking provision in an era of automation.

While these policy-based studies are inarguably important, their findings will be relevant only when a reasonable ratio of a city’s population has converted to AVS. There has been a general lack of investigation into policies that encourage travellers to switch over to driverless modes while considering the benefits

and drawbacks of automation.

In the absence of data, analytical methods that provide insight on vehicle automation in different operating scenarios can be very useful. In a recent paper, we addressed the automation controversy by using economic analysis based on supply-demand curves, capturing the impact of automation on the supplydemand equilibrium. Several insights emerged. One key conclusion is that we cannot blindly support or dispute automation without knowledge of a) the transportation infrastructure in a given area, and b) the vehicle-sharing behaviour of consumers. Based on these two factors, one of three scenarios will occur:

1. A widespread acceptance that any level of automation is beneficial for society;

2. A belief that partial automation is the best approach; or

3. A recognition that no vehicles should be automated.

Previous studies of vehicle subsidization show great promise in fleet electrification, and in our paper, we extend them by developing subsidization regimes for AVS and comparing them with two other policies founded on taxation and vehicle-sharing. We assessed the impact of three policies that can promote automation once it is deemed to be beneficial to society:

POLICY 1: Government subsidization, whereby a governing agency subsidizes AVS by offering rebates to buyers from an exogenous budget.

POLICY 2: Rebates are provided to AV owners using funds generated from a tax collected from regular vehicle owners.

POLICY 3: The industry promotes sharing AVS among groups of users to distribute ownership costs.