Introduction
Autonomous vehicles, or self-driving cars, are the transportation of the future if investments from car companies, tech companies, and governments are any indication. Most major car companies and several tech giants including Apple, Uber, and Microsoft are all working on having autonomous vehicles on the market by 2020 (Bansal and Kockelman 2016). In fact, the autonomous technology already exists and is mostly being tweaked, bug tested, and experimented with in a variety of environments. The market also appears primed for adoption (Levinson 2015). Currently most of the questions are in legal issues and what policies and infrastructure will go into making a smooth transition to the technology. These are tricky questions that will not be easy to answer, but at least in the US, state and federal governments are currently pouring resources into creating policies to ready the streets for the technology (Gonder et al. 2016).
The implications of this new technology are numerous. One of particular interest to this paper is how these vehicles will impact greenhouse gas emissions. Transportation is a key contributor to global climate change, and in 2014, 26% of US greenhouse gas emissions, the majority of which are carbon dioxide, were a product of transportation, the second largest contributor only to electricity (Sources 2014). With this big shift in transportation on the horizon, there are several theories for how self-driving cars will shift greenhouse emissions for better or for worse.
Before analyzing these theories, it is important to define autonomous vehicles, as there are varying levels of automation and implementation. For instance, most cars on the road today have some form of Level 1, function-specific automation, or Level 2, combined function automation, autonomy in the form of cruise control, emergency breaking, or self-parking capability (Levinson 2015). On the other end of the spectrum is Level 4, or fully autonomous vehicles that require no human input at all. And in between is Level 3, or limited self-driving automation, which only requires human intervention in certain limited situations (Levinson 2015). Unless otherwise stated, referring to autonomous vehicles will denote Level 3 automation, as this is what “autonomous vehicles” generally refer to, as well as being immediately forthcoming. In literature, there are also questions of how different environments will translate to different driving patterns (Wu et al. 2011). However, this paper will focus on American urban environments, as autonomous vehicles have been studied most extensively and are predicted to have the largest effects in these environments.
Literature Review
Among leading scholars in autonomous vehicle research, there is debate on what sort of effect the technology will have on carbon fuel emissions. There is consensus that the technology itself will lead to greater fuel efficiency due to more direct routing, less congestion, and communication with other vehicles (Miller and Heard 2016). However, whether or not this will translate into less fuel consumed is hotly debated and dependent on level of automation and complex travel behavior patterns. These travel behavior patterns include unknowns such as increased amounts of travel, ride sharing, travel for underserved populations, and displacement of public transportation. It is important to note that this is an example of the rebound effect, or that energy efficiency technology leads to more use of the technology, resulting in less of a decrease in net energy use than often expected (Sorrell and Dimitropoulos 2008). The magnitude of which the rebound effect negates net energy use is difficult to quantify because of different capital costs, the multitude of other variables that play a role, and the opportunity costs of time in energy production (Sorrell and Dimitropoulos 2008). All of these factors lead to difficulty in finding a single rate for the rebound effect, and, instead, the effect is entirely context dependent.
Despite all of the variables that make the rebound effect difficult to quantify, several researchers have tried and have come to similar, inconclusive results. For example, Miller and Heard claim that there is not enough information in order to properly assess the long-term carbon impacts of autonomous vehicles. They write, “The effect of AV [autonomous vehicle] adoption on consumer travel patterns may have greater influence on environmental impact than technical attributes; however, the forces associated with these behavioral dynamics are less consistently favorable” (Miller and Heard 2016). Brown et al. quantifies this by estimating the energy efficiency savings compared to constructed estimations of travel behavior patterns. They conclude that there is a potential for fuel increase of up to +173% and a potential for fuel use reduction of up to -96% if variables are maximized towards fuel efficiency or increased travel. Yet, they conclude that true effects on fuel use are unknown between that huge range of possibilities without further research (Brown et al. 2014)
While Miller and Heard, Brown et al., and most other researchers seem to be cautiously optimistic about the opportunity for reducing fuel emissions, others, such as economist Donald Dewees, are much more critical of the ability of autonomous vehicles to reduce fuel emissions. Dewees estimates potential fuel savings of $2.6 Billion/year, but only if everyone used Level 4 autonomous vehicles, which he finds to be unlikely (Dewees 2016). He continues by stating, “One individual buying an AV reaps only that insignificant fraction of the reduced congestion fuel consumption. Therefore the private incentive to buy an AV to save fuel is only a fraction of the social benefit” (Dewees 2016). Dewees sees safety as the primary private reason to switch to an autonomous vehicle, but he is skeptical that drivers will willingly give up the pleasure of driving for safety (Dewees 2016).
Other researchers have looked at how the effects of autonomous vehicles could change over time. For instance, Wadud et al. claim that it is still unknown whether the net effect on carbon emissions will be positive or negative, but it is highly dependent on level of automation. They write, “In the nearer term, at relatively low levels of automation, many of the energy intensity saving mechanisms could be realized, which would most likely outweigh the modest increases in travel activity” (Wadud et al. 2016). However, over time, especially once level 4 automation is achieved and no human intervention is required, driving habits will dramatically change resulting in increased travel and travel of underserved populations (Wadud et al. 2016). In contrast, Chen et al. predicts that there could be increased or constant fuel use at first as the technology is first deployed, but this would decrease as more automated vehicles are introduced and popularized resulting in infrastructure and further communication between vehicles that will drive fuel usage down (Chen et al. 2015).
Others have also estimated fuel savings, but through more specific frameworks. For example, Fagnant and Kockelman focus specifically on shared autonomous vehicles (SAVs), and have developed an economic model to create estimations from a series of case studies. They conclude, “Preliminary results indicate that each SAV can replace around eleven conventional vehicles, but adds up to 10% more travel distance than comparable non-SAV trips, resulting in overall beneficial emissions impacts, once fleet-efficiency changes and embodied versus in-use emissions are assessed” (Fagnant and Kockelman 2014). Other researchers including Gonder et al. and Bansal and Kockelman have built off of this research specifically on shared autonomous vehicles, and they have also come to the conclusion that shared vehicles tend to have more favorable effects on fuel efficiency in the long run. However, adoption of shared autonomous vehicles also introduces many more complex and unknown variables such as people’s willingness to carpool, abandon personal car ownership, and pay for increased capital costs (Bansal and Kockelman 2016; Levinson 2015).
Much of the recent research is now focused more on the economic policies and infrastructure that can drive a reduction in fuel emissions. Through the National Renewable Energy Laboratory, Gonder et al. uses estimates made by Brown et al. and tests them across national trends in variables such as road infrastructure, average temperatures, elevation, and road volumes (Gonder et al 2016). Gonder et al. do not arrive at a conclusion as to the effects of autonomous vehicles on greenhouse emissions, but rather look at this data for policy implications in order to encourage a reduction in fuel emissions. In addition, Chen et al. focus on fuel savings for different infrastructures. Specifically, they run case studies on automated mobility districts, such as college campuses, where only autonomous vehicles are present and communicative with each other. They conclude that there is potential to reduce fuel consumption 4% to 14% within the boundaries of the campus (Chen et al. 2016). This demonstrates how autonomous vehicles can be used in different and smaller environments, while also stressing the importance of the shared economy and communicative technology between vehicles.
Economic Analysis
The large question that all these articles center around is: how will driving patterns change due to the influx of autonomous vehicles? In order to answer this, it is helpful to look through an economic analysis of how the fuel market will change as a result of this technology. While the rebound effect is the primary factor being evaluated, there are several other factors that also play a role. Other contributing factors to increased fuel efficiency include reduced traffic congestion, smart communication, ride sharing, and reduced parking infrastructure. Additional factors that will increase fuel consumption include displacement of public transportation and unoccupied travel (Miller and Heard 2016).
As demonstrated in Figure 1, the changing demand of fuel will have a large effect on the price, quantity, and marginal user cost of fuel being produced. If fuel efficiency outweighs the increased fuel use from increased travel, then the demand for fuel will decrease, resulting in decreased price and quantity, and, ultimately, decreased greenhouse gas emissions. However, if increased travel outweighs the additional efficiency, the opposite will be true.
Figure 1: Marginal Costs of Fuel Extraction Over Time
In addition, Figure 1 also demonstrates that, if autonomous vehicle adoption is as large as it is being predicted, these effects will be amplified over time. Assuming a constant marginal cost of extraction, the total marginal cost for producing the fuel will have a small effect at first, but this will be exponentially larger as fuel companies adjust to the shift in consumer demand. One must also account for the long turnover rate of cars. A new car can generally last for at least 20 years before being replaced, so if production shifted to 100% automated cars, it would still take at least 20 years to see all the effects this technology (Fagnant and Kockelman 2015). This graphically demonstrates that even small shifts in fuel consumption can lead to large effects later on.
Yet, it is quite unlikely that this shift in fuel consumption will happen so rapidly and consistently. In all likelihood, the transition to autonomous vehicles will not produce such immediate effects, or, if it does, these effects will change as people adapt to using this technology or as the autonomous technology itself improves and changes. As described in the literature review, Wadud et al. and Chen et al. have opposing predictions for how fuel consumption will change as driving habits change, more people switch to self-driving cars, and Level 4 automation is achieved.
Figure 2: Wadud et al. Projection for Fuel Consumption
Figure 3: Chen et al. Projection for Fuel Consumption
One can see that adoption patterns will be critical for determining the environmental impact of autonomous vehicles. The economic costs and benefits will be in effect for years to come and could play out in a variety of manners. As a result, careful consideration must be taken when creating policies and infrastructure for autonomous cars to take off.
Policy Recommendations and Conclusion
Autonomous vehicles have the potential for huge environmental and economic impacts. However, as of now, the environmental impacts remain uncertain. In order to get a more accurate representation of what these impacts will look like, more research must be done, especially as the technology gets refined and more trails are conducted before autonomous vehicles are put on the market. It is also important to consider the specific context and environment that the technology is being deployed in, as it can have very different effects depending on variables such as the size, layout, transportation politics, and culture of a specific city.
It is also important to note that the costs or benefits of fuel emissions are only one small part of the total societal externalities of autonomous vehicles. Valuing the total costs and benefits of self-driving cars to society is a much more complicated process with even more uncertain variables. One large positive externality that has not yet been discussed is decreased risk of driverless cars and the benefit of that in relation to the value of a statistical life. Another large positive externality of autonomous vehicles is the increased time for leisure or work that consumers gain when not having to drive. Both the valuation of a statistical life and additional time are much easier to value through generally accepted values of statistical lives and human work hours. Overall, these benefits are estimated to create a positive value to society, and these estimations of the total benefits, including fuel use estimations, are very important to creating economic policies that accurately reflect the total costs and benefits to society.
As more research is conducted, policies surrounding autonomous vehicles can be constructed. For example, if it is determined within a certain city that there would be large environmental benefits of self-driving cars, then it might make sense to create subsidies for the producers of autonomous vehicles or tax conventional automobiles in order to incentivize the new technology. A carbon tax could also incentivize fuel efficiency technologies. On the other hand, if it is determined to be environmentally unfavorable for the city to switch to self-driving cars, it may make economic sense to tax the production or consumption of the cars. Alternatively, the city may want to further encourage and subsidize public transportation to discourage driverless cars. These policies are especially significant, as the benefits or drawbacks will compound as more and more conventional cars are replaced.
Another large question for policymakers is the question of physical infrastructure. An environmental analysis is an important faction of total societal costs and benefits and can help determine if it makes economic sense to implement separate lanes or roads for autonomous vehicles. This analysis of fuel markets can also be used to make decisions considering the decreased role of parking areas in a city with driverless cars.
A big implication of fuel consumption of autonomous vehicles is also how it will affect the switch to electric cars. For instance, if fuel consumption decreases with autonomous vehicles, it will take longer for the total marginal cost of fuel production to overcome the marginal cost of electric vehicles, resulting in moving the switch point later in time. Conversely, an increase in fuel use could incentivize a sooner switch point. However, even after the switch to electric cars, the same externalities apply to energy use as to fuel consumption.
In summary, autonomous vehicles are an exciting and uncertain new technology with huge environmental and economic considerations. Researchers, policymakers, and engineers alike will need to work together to create the best possible outcomes for our cities. Recommendations for the future would be to continue to increase federal funding for research and policy for self-driving cars as well as develop standards for liability and security. Through these recommendations, autonomous vehicles may be able reduce fuel emissions and greatly benefit society.
Works Cited
Bansal, Prateek, and Kara M. Kockelman. 2016. “Forecasting Americans’ Long-Term Adoption of Connected and Autonomous Vehicle Technologies.” In . /view.aspx?id=1392676.
Brown, Austin, Jeffrey Gonder, and Brittany Repac. 2014. “An Analysis of Possible Energy Impacts of Automated Vehicle.” In Road Vehicle Automation, edited by Gereon Meyer and Sven Beiker, 137–53. Lecture Notes in Mobility. Springer International Publishing.
Chen, Yuche, Stanley Young, Jeff Gonder, and Xuewei Qi. 2015. “Estimate of Fuel Consumption and GHG Emission Impact on an Automated Mobility District.” National Renewable Energy Laboratory.
Dewees, Donald N. 2016. “Are Automated Vehicles Coming at the Right Speed?” Working Paper tecipa-564. University of Toronto, Department of Economics.
Fagnant, Daniel J., and Kara M. Kockelman. 2014. “The Travel and Environmental Implications of Shared Autonomous Vehicles, Using Agent-Based Model Scenarios.” Transportation Research Part C: Emerging Technologies 40 (March): 1–13. doi:10.1016/j.trc.2013.12.001.
Fagnant, Daniel J., and Kara M. Kockelman. 2015. “Preparing a Nation for Autonomous Vehicles: Opportunities, Barriers and Policy Recommendations.” Transportation Research: Part A: Policy and Practice 77 (July): 167–81. doi: 10.1016/j.tra.2015.04.003
Gonder, Jeff, Yuche Chen, Mike Lammert, and Eric Wood. 2016. “Assessing the Energy Impact of Connected and Automated Vehicle (CAV) Technologies.” presented at the SAE 2016 Government/Industry Meeting, January 21.
Levinson, David. 2015. “Climbing Mount Next: The Effects of Autonomous Vehicles on Society.” doi:http://nexus.umn.edu/Papers/MountNext.pdf.
Miller, Shelie, and Brent R. Heard. 2016. “The Environmental Impact of Autonomous Vehicles Depends on Adoption Patterns.” Environmental Science and Technology 50 (12): 6119–21. doi:10.1021/acs.est.6b02490.
Sorrell, Steve, and John Dimitropoulos. 2008. “The Rebound Effect: Microeconomic Definitions, Limitations and Extensions.” Ecological Economics 65 (3): 636–49. doi:10.1016/j.ecolecon.2007.08.013.
“Sources of Greenhouse Gas Emissions.” 2014. Overviews and Factsheets. US Environmental Protection Agency. Accessed November 28.
Wadud, Z., D. MacKenzie, and P. Leiby. 2016. “Help or Hindrance? The Travel, Energy and Carbon Impact of Highly Automated Vehicles.” Transportation Research Part A: Policy and Practice 86 (10.1016/j.tra.2015.12.001): 1–18.
Wu, Changxu, Guozhen Zhao, and Bo Ou. 2011. “A Fuel Economy Optimization System with Applications in Vehicles with Human Drivers and Autonomous Vehicles.” Transportation Research: Part D: Transport and Environment 16 (7): 515–24. doi:10.1016/j.trd.2011.06.002