Autonomous Vehicles


cliff winston, is the Searle Freedom Trust senior fellow at the Brookings Institution. He is the co-author of Autonomous Vehicles: The Road to Economic Growth?, which presents an empirical analysis of the potential economic impact of autonomous vehicles.

illustrations by wacso

Published January 24, 2021


Driverless vehicles are poised to disrupt our economy and society. Autonomous vehicles will change how, when, where and why we travel, alter the design of our physical environment such as streets and shopping centers, and affect where we live and work. There is legitimate concern, of course, whether the technology will work safely and reliably from the get-go. And even if it does work as promised, some question whether it will lead to unintended consequences — as in more congestion and pollution, fewer jobs for drivers and a loss in government revenues dedicated to highway maintenance. Such skepticism about a new transportation technology is not surprising. After all, the very idea of a commercial airline industry was derided at as late as the 1920s. And when the industry made it off the ground in the 1930s, some observers asserted that its problematic technology would never improve. Today, the popular press seems similarly inclined to feed negative views about autonomous vehicles — for example, a New York Times op-ed titled “Cars Are Death Machines. Self-Driving Tech Won’t Change That.” While fear of disruptive change is understandable, such skepticism is dismaying because it diverts attention from the potentially enormous societal benefits of autonomous vehicles and from holding policymakers responsible for facilitating the adoption of those vehicles as quickly, safely and effectively as possible. I focus on those issues here.

Autonomous Vehicles 101

An autonomous vehicle drives itself by using a combination of onboard technologies to measure the distance of the car from various objects, including pedestrians, bicyclists and other cars. GPS, supplemented with highly detailed up-to-date digital maps, locates vehicles and indicates which has the right of way. Communications among AVs, as well as between AVs and roadway infrastructure, determine the location and intention of other vehicles, the condition of the roadway, and the status of traffic signals. Cooperative automated driving, in which multiple (or even all) vehicles are effectively plugged into a local network, has the potential to smooth traffic flow, improve travel time and its reliability, and virtually eliminate accidents along with loss of life and damage to property.

Innovators have been working away at the technology for many years. Automakers, tech companies and research universities are conducting road tests and simulations to improve AVs to meet the challenges of responding safely to real-world driving conditions. Indeed, the technologies are advancing at such a rapid clip that important breakthroughs occur regularly. And while the skeptics are still … skeptical, there’s good reason to believe that the autonomous vehicle industry will solve its remaining technological challenges.

AVs will spare their users the wasted time and stress of navigating through traffic and the extensive costs of accidents.

The economics of autonomous vehicles are a win-win for the private sector and consumers. Today, some nonautonomous vehicles are rented, but most are bought or leased long term and spend most of their existence in park. At this point, it is not clear if producers will sell AVs only to commercial fleets such as delivery services. In any case, consumers would benefit from having autonomous vehicles as they would get virtually all the benefits of owning but pay much less in capital and insurance costs because of efficiencies achieved in operating AVs in networks. Along with the efficiencies associated with sharing, AVs will spare their users the wasted time and stress of navigating through traffic and the extensive costs of accidents.

Getting From Here to There

When will the public be able to travel in autonomous vehicles? There are three key preparatory steps needed. First, new legislation must affirm federal responsibility for the safety of AVs and prevent states from instituting their own vehicle regulations beyond licensing and registration. Although the National Highway Traffic Safety Administration could short-circuit this process by initiating motor-vehicle-safety rule makings without additional federal legislation, it has not done so and remains uninvolved. Enabling legislation was proposed in 2018, but not passed.

Second, automakers must test their vehicles in advance of the NHTSA’s imposing vehicle safety and performance regulations through the federally authorized Federal Motor Vehicle Safety Standards. This step could take several years as NHTSA officials observe automakers’ testing and improvements in the technology and write the rules, as is standard, in consultation with industry participants. Producers of AVs that meet the standards would then be free to make their vehicles available for public use.

Elon Musk, who seems never at a loss for optimistic words on Tesla’s technology, has implied that this testing period could be substantially shorter than generally expected because Tesla vehicles are already very close to full autonomy. However, that claim is disputed by more cautious industry observers, and Tesla would probably encounter resistance from the government if travelers attempted to use its vehicles without hands-on control.

Third, once autonomous vehicles are available to the public, consumers can begin the (initially voluntary) transition from driving conventional vehicles to traveling in shared autonomous vehicles. One vision is that government would make the requisite changes to the infrastructure to accommodate autonomous travel. A more skeptical view taken by AV developers is that they are building their vehicles for today’s roads, assuming no special accommodating infrastructure other than the roads being in a good state of repair.

This transition could take many years because widespread adoption can’t happen overnight — managing the depreciation of the existing conventional fleet of some 300 million vehicles in the United States alone is one among several issues — and even longer if the technology can’t manage well without much better road infrastructure. All told, I expect that it will take at least 15 to 20 years before the public’s adoption of autonomous vehicles exceeds 50 percent, and several more years for complete adoption of shared autonomous vehicles throughout the country.

In the meantime, AVs are being assessed in U.S. cities and states that have approved testing in the absence of federal action, as well as in other countries. Autonomous shuttles are being appraised in pilot projects on closed loops in various counties. And they are already being employed in one specialized mission: driverless trucks have responded to an immediate need during the coronavirus pandemic by moving medical supplies in risky settings without much human interaction.

California has become the epicenter for developing and testing autonomous trucks. And, TuSimple, a self-driving trucking startup, plans to build a coast-to-coast autonomous freight network in the coming years.

The Benefits of Av — a Closer Look

Autonomous vehicles’ most important economic impact is likely to be seen in travel speed and reliability. This would result from smoother traffic flow as vehicles travel closer together at higher speeds and as accidents caused by operator error, which can back up rush hour traffic for hours, are virtually eliminated. Note, too, that unlike drivers in nonautonomous vehicles, the software managing AVs will not be tempted to compound traffic slowdowns by rubbernecking past wreckage.

An improvement in the average time that it takes to get from Point A to Point B and a reduction in the variation in travel time — will the commute take 30 minutes or 90 minutes? — are likely to generate efficiency gains beyond the transportation sector, benefiting urban areas that have enjoyed a renaissance in the past few decades but are now facing limits linked to traffic and parking woes. Shorter and more reliable commute times would expand individuals’ choices of employers and employers’ choices of workers, adding a welcome increase in potential competition in local labor markets. And, of course, the cost of moving goods would fall, raising productivity, increasing the flexibility of businesses to buy from the best or cheapest suppliers and lowering costs by reducing inventories.

The widespread adoption of AVs could raise the nation’s annual growth rate by at least one percentage point, in an economy that today struggles to exceed 2 percent growth.

How much of a difference will all this make in quantitative terms? At this early stage, no one can know for certain. However, Quentin Karpilow and I have made some preliminary estimates, the details of which can be found in our new book Autonomous Vehicles: The Road to Economic Growth? And they are remarkably high: the widespread adoption of AVs could raise the nation’s annual growth rate by at least one percentage point, a true breakthrough in an economy that today struggles to exceed 2 percent growth. Intuitively, the large reductions in travel costs created by AVs greatly increase the value of the highway network by exponentially increasing access to the nation’s human and physical resources. As a result, more people would be working, shopping, trading and producing goods.

Along with spurring economic growth, autonomous vehicles could produce a host of other benefits, including improvements in public health, better access to leisure activities, more efficient land use and even assistance in addressing the two leading social problems at this moment: the pandemic and violent police confrontations.

Public Health

AVs have the potential to greatly improve public health by making highway travel far safer and by liberating drivers from the stress of driving in congested conditions. According to NHTSA, the social costs from motor-vehicle crashes approach $1 trillion annually, accounting for the loss of life of some 40,000 people, injuries to roughly 4 million, and economic losses from 25 million vehicles damaged.

Surveys indicate that commuting is the leading daily activity for which individuals’ dominant attitude is negative. Indeed, automobile commuting in congested conditions can damage emotional health by causing stress that leads to road rage and household violence. AVs could also generate substantial intangible gains in health by converting stressed-out commuters to more serene passengers.

Leisure in Vehicles

Manufacturers will have incentives to change the interiors of driverless vehicles to facilitate a variety of activities — among them, napping and dining. More important, autonomous vehicles will enable people who cannot drive because of age, infirmity or lack of a driver’s license to shop, visit friends and dine out without threatening public safety.

AVs would weaken the justification for restrictive land-use policies such as parking requirements that aim to reduce congestion but which, in practice, have little effect on congestion.

Land Use

Currently, the United States has more parking spaces than cars. Since people would generally share, not own, autonomous vehicles, valuable land that was once used for parking cars in central business districts and for home garages would be freed up to, say, build new housing or expand existing homes. The reduced parking footprint for suburban malls could also make land available for housing or other beneficial uses because AVs would leave passengers at designated drop-off zones and would then be directed to parking facilities where they could be stacked vertically and easily retrieved when summoned.

Consider, too, that AVs would weaken the justification for restrictive land-use policies such as parking requirements that aim to reduce congestion but which, in practice, have little effect on congestion and raise housing prices by reducing the quantity of housing available.


While we hope Covid-19 will be only a sad memory by the time AVs have a serious presence on the road, the threat of new pandemics will persist. In this context, AVs could prove invaluable. Social distancing has been the most effective tool to limit the spread of the coronavirus. Hence, the Centers for Disease Control and Prevention has encouraged people to drive alone because carpooling, ridesharing and mass transit almost certainly increase the risk of contagion.

In a world with easy access to AV services, virtually all Americans, including the millions who currently do not keep cars, would have access to sanitized vehicles for their sole use at manageable cost. This would particularly benefit lower-paid essential service workers who may not be able to afford car ownership and cannot work from home.

Meanwhile, freight of all sorts could be delivered in driverless trucks, sparing delivery workers (and goods recipients) from contact. Small autonomous vans have already been used in China to deliver medical equipment to first responders. In the United States, Nuro, a robotics company, has won approval to transport pharmaceuticals via AVs from CVS pharmacies to customers in Houston. More generally, by reducing disruptions to the workforce and the complex supply chains on which most production now depends, AVs would help maintain economic activity without threatening public health.

Safe and Efficient Policing

The United States has more than 800,000 state and local law enforcement officers, and a large fraction of their work consists of enforcing traffic laws. They annually pull over some 20 million motorists, respond to 6 million car accidents and, of course, document these efforts by filling out endless forms. On average, more than one officer per week is killed in a highway accident, accounting for nearly one-third of all police officer deaths in the line of duty.

It’s also worth stressing that AVs would eliminate police stops and potential violent confrontations with vehicle occupants.

AVs could virtually eliminate the need to use the police to enforce traffic laws, either sparing taxpayers the expense or allowing for police redeployment to more pressing concerns. After all, AVs will be programmed to obey traffic laws, to eliminate driving errors and to drop off passengers where it is safe and legal. What’s more, autonomous vehicles don’t drive under the influence of intoxicants and don’t become fatigued.

It’s also worth stressing that AVs would eliminate police stops and potential violent confrontations with vehicle occupants. Confrontations involving African-American drivers in nonautonomous vehicles have all too frequently led to fatal police shootings and ever greater fear and frustration in communities of color. This deadly cycle would be minimized in a world of driverless vehicles, sparing the nation both misery and shame.

What Must Be Done

AVs will operate on a publicly owned and managed road system. Hence, policymakers and regulators have a critical (and unavoidable) role to play to ensure that society will realize the potential benefits from those vehicles. Specifically, policymakers must establish a framework for vehicle testing and adoption, invest and modernize the highway infrastructure to facilitate safe vehicle operations and reform highway policies to encourage efficient operations.

Testing Framework

In 2018, Congress drafted, but failed to pass, important AV legislation that would have clarified and expedited national vehicle testing and adoption. Two years later, Congress has still not passed a version of this legislation. As noted, states and localities have adopted regulations allowing testing within their borders, sometimes with an eye toward attracting new high-tech businesses. However, federal legislation is necessary to jump-start the formal adoption process.

Government Investments

In a fully optimized system, AVs would be connected to other vehicles and their surroundings, including pedestrians, infrastructure and the road network.

Some states are making investments to upgrade their infrastructure to align it in advance with AV technology. For example, Panasonic and Utah’s Department of Transportation are jointly developing a system that enables vehicles, roads and traffic signals to continuously communicate about conditions, location of obstacles, signal timing and the speed, direction and position of all cars in the network. Michigan, for its part, is planning to work with a Google-funded startup to transform a stretch between Detroit and Ann Arbor into a connected highway for testing purposes.

Several states are installing fiberoptic lines in roads that can send electronic warnings to AVs about hazards ahead and other information to keep them aware of their surroundings.

By contrast, other states seem content to wait for the feds to provide guidance before moving forward. Regardless of who is at fault, states that delay upgrades to their infrastructure could delay the use of autonomous vehicles on their roads.

Reform Highway Policies

AVs are not immune to highway inefficiencies. Their operation could be disrupted by congestion and substandard roads that require vehicles to slow down or change lanes. As noted, we can expect AVs to reduce travel delays, but even great technology won’t be able to mitigate congestion beyond a certain level of road usage.

To complement their efficient operation, we need to adopt congestion pricing, charging tolls transparently linked to the societal cost of adding one more vehicle to the road and giving travelers route options before they depart. By the same token, we need to tie broader road use charges to the wear and tear that specific types of vehicles cause — in particular setting charges that encourage truckers to reduce their weight per axle or to shift to trucks with more axles and thereby reduce road damage.

This latter reform is worthy in itself. But it is of particular importance in a transition to AVs because road quality will have a lot to do with managing the smooth flow of densely packed vehicles and with preserving the economic gains from faster and more reliable travel.

What Could Possibly Go Wrong?

The most important concern about autonomous vehicles is the most obvious: that the technology will not be up to the task of operating under real-world conditions. Other concerns are that they will increase congestion, pollution and sprawl; reduce employment; exacerbate highway and transit deficits; and create accident liability issues. Let’s look at them individually.

Technology. Skeptics like to point to the handful of accidents that have occurred during tests of the vehicles and to unresolved technological challenges that could expose humans to danger in difficult driving conditions. I am far less concerned — at least in the long run — for two reasons. First, AVs are being developed competitively by various combinations of all the world’s leading automobile and technology companies. They are not the brainchild of one or two firms trying to disrupt an established industry, fend off incumbents and initially satisfy niche consumers. It will take time, but all car companies are signaling that they will eventually abandon their nonautonomous vehicle production. Accordingly, they have enormous incentives to innovate and to make whatever investments it takes to succeed because their survival will be at stake.

Second, as noted, the technology is advancing rapidly. Take one example: lidar (light detection and ranging), which allows AVs to see their environment in three dimensions and to avoid collisions. Indeed, lidar seems ready in terms of accuracy, detail and cost, with a half dozen firms working on it.

Highways will be able to accommodate a greater number of autonomous than non-autonomous vehicles because the former can safely travel close together at high speeds and because highways can be reconfigured with more, but narrower, lanes.

Nonetheless, critics are likely to modify their views only when autonomous vehicles are widely adopted and they have track records of operating safely and efficiently. There are, of course, risks in moving into virgin territory. At the same time, there are enormous opportunity costs from delaying the adoption of AVs.

Congestion and Sprawl. Skeptics are inclined to dismiss the potential for AVs to ease congestion and sprawl. They reason that if AVs do indeed lower the true cost of auto travel (including wasted time), they are likely to induce more automobile travel and to create sprawl by making it more practical for people to live in less-expensive homes farther from their workplaces.

However, it’s important to remember that highways will be able to accommodate a greater number of autonomous than nonautonomous vehicles because the former can safely travel close together at high speeds and because highways can be reconfigured with more, but narrower, lanes.

Congestion pricing can discourage peakperiod travel and reduce stop-and-go driving that increases pollution. And households would be encouraged to take those charges into account when they make their residential location decisions.

Jobs. There is no doubt that the widespread adoption of AVs would force hundreds of thousands of people who drive for a living to look for new work. But it’s likely that those job losses would be more than offset by additional jobs created by the demand for workers in everything from maintaining AVs to updating the very accurate maps needed for them to operate safely and, more important, by opening opportunities throughout a more efficient economy.

Highway and Transit Fiscal Deficits. Autonomous vehicles’ improved fuel economy from traveling in a smoother traffic flow with much less stop-and-go driving will reduce gasoline tax revenues that are the backbone of highway finance. Indeed, they may drive this revenue source close to zero because AVs designed from the ground up will almost certainly be battery- or hydrogen-powered. Efficient road pricing would do double duty here, eliminating highway revenue shortfalls even as it reduces congestion.

By the same token, AVs will probably increase deficits in urban mass transit by contributing to its long-run decline in ridership. However, the efficient action is to curtail transit operations because those operations will effectively be replaced by a personalized transit system provided by autonomous vehicles.

Liability. The legal community has just begun to confront the issue of who pays when AVs have accidents. Liability has yet to be firmly established, and, as with other new products, the law is likely to evolve as AVs are adopted and the various stakeholders — insurance companies, the tort bar and, of course, the public — gain greater experience with them.

I have two things to say here.

First, clarifying liability, regardless of how it is assigned, is important because certainty on this matter will be needed for the AV industry to thrive.

Second, liability should fade markedly as an issue of contention because AVs will be so much safer than conventional vehicles and because, like airlines, manufacturers will continue to solve their residual safety issues. Note, too, that the data automatically collected by AVs will be very useful in accident forensics and determining fault.

• • •

My optimism about what AVs are likely to do for us raises one important question: why should we expect government policy toward AVs to be enlightened when there is a long history of Rube Goldberg policymaking toward autos that has simultaneously subsidized and taxed them and given short shrift to efficient highway travel?

My answer is that, this time, the whole world will be watching as countries, cities and states compete intensely to develop and adopt autonomous vehicles.

Given the enormous benefits at stake and the visibility and importance of global, interstate and intercity competition, policymakers who delay in reforming highway policies will incur large political costs for failing to realize the astonishing promise of driverless transportation as soon as possible.

main topic: Tech & Telecoms