Itdp 3r Report Final

Three Revolutions in Urban
TRANSPORTATION
How to achieve the full potential of vehicle electrifica-
tion, automation and shared mobility in urban trans-
portation systems around the world by 2050
Lew Fulton, UC Davis

Jacob Mason, ITDP
Dominique Meroux, UC Davis
Research supported by:

ClimateWorks Foundation, William and Flora Hewlett Foundation, Barr Foundation
SUSTAINABLE TRANSPORTATION ENERGY PATHWAYS

of the Institute of Transportation Studies
Acknowledgements
The authors wish to thank ClimateWorks Foundation, William and Flora Hewlett Foundation, and Barr Foundation for
their generous financial support, which made this work possible. We also thank the individuals who participated in the
advisory committee for this study and provided valuable comments. Members of that group included Pierpaolo Cazzola
(IEA); Holger Dalkmann, Dario Hidalgo, and Diego Canales (WRI); Phillippe Crist (ITF); Shomik Raj Mehndiratta (World
Bank); Tim Wallington (Ford); Jarret Walker (unaffiliated); Anthony Eggert (ClimateWorks); Margarita Parra (Hewlett
Foundation); Colin Hughes, Aimee Gauthier, Clayton Lane, Kathleen Letchford, Ana Nassar, Bernardo Serra, Diego da
Silva, Gabriel Oliveira, and Clarisse Linke (ITDP); and Dan Sperling, Susan Pike and Mollie DAgostino (UC Davis).
Thanks also to Jacob Teter and other staff at the International Energy Agency for insights related to the urban analysis
developed with the IEA Mobility Model (MoMo) that this analysis draws on. However, this analysis has been undertaken
by the UC Davis and ITDP, and does not necessarily reflect the views of the International Energy Agency.
Finally, the authors thank all the members of the production teams at UC Davis and ITDP who carried out editing, proof
reading, and layout of this report. This includes Stephen Kulieke, Beth Bourne, Rosa Dominguez-Faus, and Kelly Chang
(UC Davis); Jamie Knapp (J Knapp Communications); and Jemilah Magnusson (ITDP).
Many other people too numerous to name helped this project succeed, and we thank them. Of course, any errors or
flaws in the end product are the responsibility of the authors alone.
i Three Revolutions in Global Transportation

Contents
Acknowledgements i
Executive Summary 1
1. Introduction 5
2. Study Design, Methodology, and Scenarios 6
3. The Three Revolutions: Status and Potential 8
Electrification 8
Automation 9
Shared Mobility 11
4. Future Scenarios: BAU, 2R, and 3R 15
The BAU Scenario 15
The 2R Scenario: Electrification and Automation 16
The 3R Scenario: Adding Shared Mobility 19
Results: Passenger Travel Projections Across the Scenarios 21
Scenario Impacts on Energy and CO2 Emissions 25
The Costs of 3 Revolutions 29
5. Policy Narratives 32
BAU Scenario Policy Narrative 32
2R Scenario Policy Narrative 33
3R Scenario Policy Narrative 34
6. Conclusions, Uncertainties, Next Steps 36
Bibliography 37
ii Three Revolutions in Global Transportation

Executive Summary
The world is on the cusp of three revolutions in by encouraging a large increase in trip sharing, transit
transportation: vehicle electrification, automation, use, and active transport through policies that support
and widespread shared mobility (sharing of vehicle compact, mixed use development, cities worldwide
trips). Separately or together, these revolutions will could save an estimated $5 trillion annually by 2050
fundamentally change urban transportation around the while improving livability and increasing the likelihood of
world over the next three decades. meeting climate change targets.
Each revolution addresses different societal needs, but

can also lead to societal costs: Methodology
Vehicle electrification can cut vehicle energy use We build on two recent reports published by ITDP and
and CO2 emissions. However, for electrification to UC Daviss STEPS program: A Global High Shift Cycling
have maximum benefits, power generation must be Scenario (2015) and A Global High Shift Scenario:
strongly shifted away from fossil fuels and deeply Impacts and Potential for More Public Transport, Walking
decarbonized. In addition, these vehicles will likely and Cycling with Lower Car Use (2014). Both reports took
remain expensive for at least one more decade. a scenario approach to consider the role of different travel
modes in providing mobility, and the amount of potential
Automation can provide important safety benefits, energy savings and CO2 reduction that could come from
reduce labor costs, and enable cheaper travel a less car-centric world in the future (Mason, Fulton, &
and more productive use of time. However, by McDonald, 2015; Replogle & Fulton, 2014).
lowering the cost of travel in terms of time and
money, automation would likely induce more travel This report expands upon the scope of the previous
and dramatically reduce the number of jobs in studies by considering the role of electrification,
transportation. automation, and ride sharing (more people per vehicle)
in developing future scenarios. The possible types of
Shared mobility, whether through shared vehicle trips impacts are well documented, and researchers have
or public transport, can lead to more efficient use of begun to estimate how various combinations of impacts
urban space, reduce traffic congestion, enable more such as people spending more time in their cars, or
walking and cycling, cut energy use and emissions, on-demand mobility trips substituting for public transport
and generally improve urban livability. However, may affect travel and energy use. But most studies have
this would require large increases in load factors not explicitly projected numeric scenarios into the future
(passengers per vehicle trip), and a range of strong or attempted to characterize how various interactions
policies to achieve. could play out. As with our previous modal shift studies,
this study is global, breaking the world into eight regions
Together, the positive and negative aspects of each including five major markets: United States, Europe, China,
revolution will interact in many complex and difficult- India, and Brazil.
to-predict ways. This report may be the first to attempt
to quantify how these major changes could evolve and We have developed our present analysis using three
interact on a global and regional basis out to 2050. It main urban travel scenarios: a business-as-usual scenario,
considers possible end states, as well as transitional a technology-dominated 2 Revolutions scenario, and a
pathways and policies needed to get there. technology + high shared-mobility 3 Revolutions scenario.
These are elaborated from a base year of 2015 through
Our central finding is that while vehicle electrification 2050 as follows:
and automation may produce potentially important
benefits, without a corresponding shift toward shared Business-as-usual (BAU) scenario This scenario
mobility and greater use of transit and active transport, assumes few changes from 2017 travel patterns and
these two revolutions could significantly increase current trends through to 2050. No major revolutions
congestion and urban sprawl, while also increasing the occur. It assumes internal combustion engine (ICE)
likelihood of missing climate change targets. In contrast, light-duty vehicles (LDVs) remain dominant or grow in
1 Three Revolutions in Global Transportation

dominance, depending on the country, through 2050, 3R Scenario Global Results
and applies population and growth projections with
these assumptions in mind.
Compared to the BAU case in 2050, the 3R scenario
2 Revolutions (2R) scenario This is a technology- produces impressive global results. It would:
focused scenario that includes rapid vehicle
electrification along with but starting later rapid Cut global energy use from urban passenger
automation. Electric vehicles (EVs) achieve a transportation by over 70%
significant share of vehicle sales by 2025 in leading Cut CO2 emissions by over 80%
countries, with automated EVs reaching this stage Cut the measured costs of vehicles,
about five years later. Both are dominant around the infrastructure, and transportation system
world by 2050. This scenario contains no significant operation by over 40%
increase of shared vehicle trips through new Achieve savings approaching $5 trillion per year
technology; it preserves the BAU trends toward a
private-car-dominated world.
savings, mostly after 2030, and only with largescale
3 Revolutions (3R) scenario This scenario decarbonization of electricity production. In the
includes widespread vehicle electrification and 2R scenario, vehicle travel rises higher than in the
automation, and adds a major shift in mobility BAU, but vehicle-related emissions and energy
patterns by maximizing the use of shared vehicle use are eventually cut significantly, with specific
trips. This scenario includes all three revolutions, and CO2 reductions dependent on the extent to which
is a strongly multi-modal scenario, with increased electricity production decarbonizes around the world.
availability of vehicles for shared trips, increased If the worlds electricity production is not completely
public transport availability and performance decarbonized by 2050, this scenario may produce
(including on-demand small bus services, larger buses more CO2 emissions in 2050 than is consistent with
and rail), and significant improvements in walking and targets to limit global temperature rise to 2C (or less)
cycling infrastructure and therefore in travel by these compared to preindustrial levels.
modes.
An autonomous vehicle (AV) world without
Other scenarios with different combinations of these electrification (i.e. using ICEs) and without trip
revolutions could be considered; the choices made sharing would not cut CO2 emissions out to 2050.
here are intended to simplify these complex scenarios We estimate that the lower travel time costs
and highlight certain features. And although we cannot provided by self-driving vehicles would likely lead to
accurately predict the interactions that each step of each a significant increase in vehicle travel, on the order of
revolution will have on the others, our scenarios create 15-20% compared to the BAU (with a wide range of
paradigms of travel that we can use to quantify the energy uncertainty). The increased efficiency of AVs would
and CO2 impacts and begin to develop policies to guide offset some or all of this travel to keep energy and
the world toward the most societally optimal outcomes. CO2 close to BAU levels; but it is the widespread use
of electrification in AVs that dramatically reduces
vehicle-related pollution and CO2 emissions in this
Findings scenario. The increased travel of AVs could trigger
more traffic congestion, though their improved road-
Our central finding is that the 3R scenario is the best space efficiencies and coordinated travel patterns
option for reducing energy use and CO2, and performs might mitigate some of these impacts. We do not
significantly better than 2R in these respects as well attempt to estimate congestion impacts in this study.
as on total measured cost. The 3R scenario would also
dramatically reduce the number of vehicles on the The 3R scenario performs significantly better
worlds roads. This finding is true worldwide and for each on energy and CO2 , as well as on livability. This
individual country or region studied. scenario has the potential to deliver an efficient,
low-traffic, low-energy, and low-CO2 urban transport
The following summarizes all key findings: system around the world. In this scenario, the
widespread adoption of on-demand travel with
The 2R scenario, which includes electrification substantial ride sharing, along with greater use of
and automation but with a private-car-dominated (high-quality) public transport, cycling, and walking
world, may provide significant energy and CO 2 reduces car travel by well over half in 2050, and the

number of cars by nearly three-quarters compared
to our BAU. It would reduce traffic congestion and * Lower overall travel demand due to shorter travel
parking needs dramatically, opening up tremendous distances from more compact cities
amounts of urban space for walking, cycling, and * More transportation choices, with walking and
other uses. This scenario with energy use and CO2 cycling rising significantly over time given safer
emissions in 2050 less than one-third of the BAU and conditions and better infrastructure
about one-half that of the 2R scenario, and with fully * A much larger share of travel provided by more
decarbonized electricity production yields a very efficient modes (bus and rail systems as well as
low CO2 picture worldwide. smaller, right-sized vehicles, whose size
better matches travel demand)
Ride sharing must deliver high-occupancy-vehicle * A higher average load factor (people per trip)
travel, both in light-duty taxi-style vehicles and in * More intense vehicle use, requiring far fewer
some larger vehicles such as minivans and small vehicles to meet passenger travel needs (since
buses. Ride hailing services do not help bring about personal vehicles currently remain idle 90+% of
this scenario if they are dominated by single-occupant the time)
trips. Thus, we distinguish ride hailing from ride * Lower parking and road-building requirements
sharing where the latter means separate trips are from less vehicle travel), with associated cost
shared in a single ride. Our assumed load factors savings.
(average passengers per trip) in ride-hailed vehicles
rises over time in the 3R scenario and is about 30- The 3R scenario would also dramatically reduce
40% higher than in the 2R or BAU scenarios by 2050. the number of vehicles on the worlds roads by
2050. The current global urban stock of LDVs,
The 3R scenario also delivers large cost savings. around 750 million (out of 1.1 billion total, urban +
The costs of urban travel would likely be much lower non-urban), reaches 2.1 billion by 2050 in our BAU
overall in the 3R scenario than in the 2R or BAU and 2R scenarios. In the 3R scenario it drops instead,
scenarios, considering a wide range of out-of-pocket to about 500 million. In 3R these far fewer vehicles
costs, including vehicle purchase and operation, are highly productive, carrying many more people on
fuel purchase, the costs of operating transportation more trips per day than average vehicles in the other
network companies (TNCs) as well as public transport scenarios. Fewer vehicles in 3R allows the world to
systems, and the costs of building and maintaining build far fewer parking spaces and lots, and frees
road and transport infrastructure. These savings up considerable space for other activities. Total LDV
emerge mainly after 2030 and relate mainly to lower travel also drops by half in the 3R scenario compared
costs of vehicle purchase (given far fewer vehicles to the 2R scenario, meaning less congested and
purchased), energy cost savings, and road and safer roads. In turn, the urban landscape can be
parking infrastructure cost savings. The 2R scenario repurposed and reoriented toward more cycling and
saves some costs by 2050 compared to BAU from walking.
lower-cost EV and AV operation and by eliminating
most drivers, but these savings are mostly offset by Our findings are broadly consistent across world
higher cost vehicles and induced, increased travel. regions, despite very different starting points.
Since countries like the United States, with its car-
Other potentially important benefits are more dominated transportation system, are very different
difficult to quantify. Though not specifically from, say, India, with its wide range of modes sharing
calculated, the value of CO2 and criteria pollutant the streets, these scenarios also look quite different.
emissions reductions are potentially important in In fact, for India and most other emerging economies,
the 2R scenario, along with the value of congestion the high levels of shared vehicle trips in the 3R
reduction in the 3R scenario. Both scenarios should scenario (at least in terms of the dominance of mass
provide substantial safety benefits if automation lives transit mobility) already largely exists, and the main
up to its safety potential and given the much less car- challenge is to preserve it. In general, those regions
dominant world in 3R. Quantifying these impacts is an with existing high levels of public transport, walking,
important area for further research. and cycling see these travel modes decline in the
BAU scenario as well as in the 2R scenario, as cars
The 3R scenario achieves its energy, CO2 and become dominant everywhere; in contrast, they
cost savings by creating a far more efficient retain or gain on public transport ridership in the 3R
transportation system than in the BAU or 2R scenario, thanks to major investments in systems and
scenarios, including: strong linkages with shared vehicle trips. A 50% or

greater reduction in both the numbers of vehicles and contain as a core policy on-demand ride sharing
CO2 emissions in 2050 appears possible everywhere incentives, such as vehicle travel fees tied to vehicle
in the 3R scenario vs. the BAU scenario. occupancy (such fees could also be applied to private
vehicles). These could also vary with the length of
Policy Implications of the Scenarios trips, vehicle carbon intensity, and level of congestion.
Policies could also restrict or heavily charge for
The intensity of policies likely required to achieve each private ownership of AVs, and/or their undertaking of
scenario tends to increase moving from the BAU case zero-occupant trips. Incentives would urge a better
to the 2R and 3R scenarios, with the latter scenario match between vehicle size and occupancy, which will
requiring the most ambitious policy scheme to achieve make travel more efficient. Bicycle and e-bike sharing
the maximum societal benefits. All policies envisioned systems would be encouraged. Multimodal urban
in these scenarios would require some flexibility and planning investments in walking, cycling, and public
iteration, as it cannot be fully known how various factors transit infrastructure and services would likely help
will interact with each other. They would also probably reduce and shorten vehicle trips. These investments
vary by country given local conditions and preferred would be most effective if coupled with proven
policy strategies. The following summarizes our policy practices such as implementation of compact, mixed-
needs assessment: use urban plans centered on linking concentrated
development zones through public transit, and
The 2R scenario will require a dual-policy focus featuring designated cycling and walking zones.
incentivizing EV uptake and enabling automation.
The scenario includes strong, proactive vehicle In 3R, governments play a central role. The 3R
electrification policy incentives, resulting in the scenario may need governments to coordinate both
widespread adoption of EVs with steady 30% or AV infrastructure and management of public and
more annual increases in these vehicles for the next private trips, broadening the definition of publicly
20 years and beyond, and particularly rapid growth funded transportation in favor of seamless regional
between 2020 and 2035. Enabling policies include travel networks. Central to this transition is the
ongoing purchase incentives and public awareness government role for filling gaps, and maintaining
campaigns, strong government coordination and equitable access and mobility for all individuals,
support of expanded EV charging infrastructure, regardless of income, disability or access to a
research support, elimination of petroleum subsidies, smartphone or vehicle.
as well as electricity decarbonization policies such as
carbon taxes or cap-and-trade systems. Overall, this analysis suggests that a combination of
electrification, automation, and multimodal shared vehicle
Achieving widespread driverless vehicles must trips would bring by far the greatest societal benefits
focus on barrier removal. The commercial vehicle for every country in this study. But achieving the full 3R
sector is eager to take advantage of the tremendous scenario will require unprecedented levels of policy
cost-saving potential of automation technologies. support; it will require creativity and vigilance to ensure
Therefore, the 2R scenario assumes that a relaxed that not one or two, but all three, revolutions move
regulatory environment will enable rapid adoption of forward and to prepare cities around the world for a new
driverless cars. This differs from the BAU scenario, era of travel.
where there is an assumption that heavy regulatory
burdens delay adoption of AVs. In the 2R scenario
the policy climate is favorable enough to ensure a
widespread uptake of automation by commercial
fleets by 2025, with households following suit shortly
thereafter. But in 2R, on-demand mobility does not
grow, and preferences for private vehicle ownership
and solo driving endure, with longer trips and even
zero-occupant driving becoming commonplace.
The 3R scenario will require strong additional

support for ride sharing, public and active transport,
and land-use planning that helps to shorten most
vehicle trips. In addition to including those policies
assumed for the 2R scenario, the 3R scenario would

1. Introduction
During the 20th century, several revolutions occurred in never fully materialize. They may go in different directions
transportation systems around the world most notably and interact in unpredictable ways. They could lead to
the internal combustion engine (ICE), mass production heaven or hell scenarios, depending on the impacts
of automobiles, high speed urban and interurban rail on travel, traffic congestion, safety, energy use, and
systems, and construction of major roadway and limited emissions that result from their combined uptake (Chase,
access expressway networks. However, in the latter part 2014).
of the century innovation slowed. Now, in 2017, most
people still move around cities primarily in vehicles with Despite the uncertainty, there is a growing literature on
ICE gasoline or diesel engines, always with a driver, and these phenomena individually, and an emerging one on
often with the only occupant serving as driver. their potential interactions (Anair, 2017; Cohen & Shirazi,
2017; Handy, 2017; Ory, 2017; Polzin, 2017). However, this
This report will discuss how the following three is one of the first studies that attempts to craft coherent
advancements in technology are set to make dramatic future transitional scenarios. Via three scenarios we
changes: explore the potential impacts of the three revolutions
on travel patterns, vehicle sales and stocks, energy use,
Electrification After an initial surge and rapid CO2 emissions, and costs. We undertake this examination
decline in the late 19th century, the electric vehicle on a worldwide basis and with individual results for key
(EV), either hybridized with engines or entirely running countries and regions.
on batteries, has re-emerged as a viable technology.
By early 2017, 2 million electric and plug-in hybrid Many other scenarios could be developed, but the
passenger cars (and 200 million 2-wheelers) are specific ones chosen for this report are designed to
plying the worlds roads (Lutsey, 2017). highlight a technology dominated world with and without
an additional revolution in the travel patterns the world
Automation Although they are not commercial adopts.
yet, technologies to automate vehicles, eventually
including eliminating the need for drivers, are moving
rapidly. U.S. Society of Automotive Engineers (SAE)
level 4 driverless cars appear on track to begin
entering commercial fleets by the early 2020s (SAE,
2016).
Shared vehicle trips While public transport

systems emerged around two centuries ago, and
taxi services and carpooling have allowed people to
share trips for decades, new technology creates the
potential for nearly all trips to be easily shared among
multiple riders. This development could revolutionize
transportation. Cutting the cost of ride hailing in half
or more, ride sharing has the potential to attract
large numbers of travelers and dramatically cut the
numbers of vehicles on the worlds roads. However,
these benefits are only significant if they reduce the
number of trips taken. Taxi services or transportation
network companies (TNCs) that are not shared do not
reduce trips.
These three revolutions are highly uncertain in many

ways. They may happen soon, take decades to mature, or

2. Study Design, Methodology, and Scenarios
This report builds on two previous urban travel studies As we did in the previous two High Shift studies in this
undertaken by UC Davis and ITDP: A Global High series, we utilize basic data and projections from the
Shift Cycling Scenario (2015) and A Global High Shift Mobility Model (MoMo) created by the International
Scenario: Impacts and Potential for More Public Transport, Energy Agency (IEA, 2016), though we have created a new
Walking and Cycling with Lower Car Use (2014). These spreadsheet tool specific to the current study, focused
reports developed an approach of comparing a business- on modeling the three revolutions and considering
as-usual (BAU) future to one that considers dramatically scenarios where they are combined in different manners.
increasing investments into sustainable transportation Our projection system tracks the numbers of vehicles
infrastructure along with other measures to promote of all major types, their cost, daily and annual travel,
efficient modes, namely public transport, cycling, and average passenger loadings, fuel use, and CO2 emissions
walking. These High Shift studies envisioned cities that worldwide, broken into eight countries and regions.
are far less car-dependent, energy- and CO2-intensive,
and perhaps surprisingly far less expensive for These variables are linked using basic arithmetic
society. relationships that allow, for example, the calculation of
total vehicle and passenger kilometers for each mode
This study uses the same basic methodology as the and summed across modes, total numbers of vehicles
previous studies, including the development of a BAU in use, total vehicle and passenger kilometers (pkm) of
scenario and several high shift scenarios. However, here travel, total fuel use and CO2 emissions, etc. The data
we are considering a more complex set of dynamics have generally been validated for 2015, but in some
than in the previous studies, since this includes an effort cases broad assumptions are made, such as the average
to understand the combined impacts of three separate passenger loadings in buses in parts of the world. We
revolutions. And although this approach is built upon the track the stocks of all vehicles but only track sales of light-
foundation of the other studies, it is worthy of a different duty vehicles (LDVs), both household and commercial
title than the other studies, hence our 3 Revolutions LDVs, and how changes in the number of vehicles
frame. needed to deliver the transportation service (measured
in pkm) translates into sales, taking into account vehicle
This study specifically considers the following: usage rates and lifetimes, such as the faster turnover of
commercial vehicles given their much more intensive daily
Explicit pathways for increased electrification in all use than household vehicles.
types of urban vehicles and modes, and plausible
time frames for a revolution electrification case in The analysis is conducted across eight countries and
each world region. world regions: United States, Organization for Economic
Cooperation and Development (OECD) Europe, China,
Plausible pathways and time frames for the India, Brazil, Other Americas, Africa/Middle East, and
introduction of vehicle automation, and assumptions Other Europe/Asia. The particulars of each region are
about the characteristics of these vehicles and how reflected in the initial data and BAU projections for each.
they may change over time. For example, some regions, such as India, already have
very high levels of shared vehicle trips, including public
The potential for increased trip sharing outside of transport use. Others, like the United States, have very
public transport systems, and better integration low levels of shared vehicle trips, while Europe falls in the
between smaller vehicles and public transport middle. Our results bring out some of these particularities
vehicles such as small and large buses, and trunk of each region, though in this main report we focus on
metro, bus rapid transit, and light rail systems. global totals and offer some examples from different
countries. We plan to produce separate materials with
Levels of public transport ridership, cycling, and more details on our results by country/region.
walking in our 3 Revolutions case that are consistent
with those in our High Shift reports, though the
massive expansion of shared trips in smaller
commercial vehicles leads to some adjustments in
public transport, cycling, and walking levels.
In this report, we consider three main scenarios:
Business as Usual (BAU) Current trends continue 3 Revolutions (3R) Here, the third revolution, an
without any revolutions. ICE vehicles remain dominant increase in shared vehicle trips, is overlaid on the first
through 2050. All vehicle trips continue to require two. We view this revolution in an expansive sense:
drivers. In countries like the United States, 85% private vehicles replaced with ride hailing of TNC
of trips remain in cars (of increasing size), most vehicles, shared vehicle trips leading to much higher
frequently with a single occupant. In other countries, average vehicle occupancy, and all this coupled
public transport shares decline as ridership grows with a strong role for public transport and active
only slowly while car ownership and travel steadily travel. These all fit together well since the world with
rise. This scenario may or may not be likely, but in any more shared vehicle trips will see vastly fewer cars
case, it provides a useful basis for comparison with and open up an enormous amount of urban space
the other three scenarios. for things like walking and cycling. Of the three
revolutions, widespread shared mobility may be the
2 Revolutions (2R) In this scenario, we consider most challenging to achieve and most dependent on
electrification and automation. It is natural to think strongly supportive policies. The potential for getting
that these revolutions will co-evolve because of their large numbers of people to share rides is highly
co-benefits; electric autonomous vehicles (AVs) can uncertain, especially as travel costs drop from the
recharge themselves easily at convenient times, and other two revolutions. Strong financial incentives will
EVs can easily supply power to the hardware needed likely be needed to encourage trip sharing and use of
to automate vehicles. EVs can also help to lower the public transport in the face of otherwise cheap point-
per-kilometer travel cost of high-use AVs. Electric, to-point services in single-occupant services.
driverless vehicles likely will be expensive to produce,
at least over the next 10-20 years, but inexpensive We also briefly consider the potential impacts of the
to operate both privately and commercially. It is revolutions individually, and as suboptimized versions of
certainly possible to have one without the other (in the main scenarios.
either direction), but together they provide a true
transformative technological and travel revolution.
Synergies Achieved by Combining Revolutions
Much of the analysis in this paper hinges on the types of synergies that could occur by combining these
revolutions. Some of these synergies are listed below (Anair, 2017):
Electrification can assist in the power and electronic demands of AVs.

Automation can assist electrification in terms of battery operation and recharging management, such
as automatically seeking opportunities to recharge during slow periods.
Similarly, AVs can help manage recharging of shared vehicles between trips and extend their effective
daily driving range in this manner.
Automation can lower the costs of sharing vehicle trips including public transport services by
eliminating driver costs, which can be 50% or more of ride-hailing costs. However, this also could lower
the costs of non-shared ride-hailing trips enough that there is less incentive to share trips or even to
take public transport.
Trip sharing and strong public transport can help overcome the tendency of automation to trigger
increases in travel, as consumers will pay for trips at the margin, and may continue to budget their time
spent in travel in a similar way as they do today (rather than purchase more comfortable vehicles and
spend more time in them).
Widespread trip sharing and use of public transport can cut the number of vehicles in use dramatically
and reduce traffic levels and congestion significantly, and (on a societal basis) provide cost savings that
more than offset the higher purchase costs of automated EVs.

3. The Three Revolutions: Status and Potential
This section provides a brief introduction and status report cost, greater time to charge an electric battery, a lack of
on each of the three revolutions, taking into account public charging stations, and a recent decline in oil prices
technology development, extent of market development, worldwide.
and various barriers that the revolution faces to achieving
large scale adoption. Increased driving range would enable more effective
deployment of EVs, especially in car share and ride share
Electrification fleets, because range is inversely related to frequency of
needing to pull a vehicle out of operation to recharge. The
Electric vehicles have already arrived: in 2016, the number recent and expected future trend in EV models is in this
of different (4-wheel) models available in countries around direction. Among the many examples, the 2017 BMW i3
the world exceeded 100, including everything from EV with a 94 Amp-hour battery offers 114 miles of electric
electric minicars to plug-in hybrid sport-utility vehicles range, up from 81 miles in the 2016 model. Notably,
(Fulton, Seleem, Boshell, Salgado, & Saygin, 2017). In the 2017 Chevrolet Bolt EV offers 238 miles of range, a
2017, several 200+ mile (300 kilometer) all-electric models substantial improvement over the 82-mile range of the
will be introduced, possibly ushering a new era of higher smaller 2016 Chevrolet Spark EV.
range EVs, and reducing what has been a major barrier to
widespread adoption. And yet the global market share of Charging Infrastructure
EVs is less than 1%. Are they succeeding or failing? Here
we briefly review their status and consider their future Greater density of charging infrastructure makes EVs a
potential. more viable option. More specifically, availability of fast-
charging stations is necessary for car share and even
Global Electric Vehicle Market more so for ride share fleets to minimize recharging
downtime. Car share and ride share fleets have lower
Varying levels of vehicle electrification, including handling costs if their vehicles are closer to EV charging.
traditional hybrid vehicles (that dont plug in), have
become available across popular vehicle platforms BMW ReachNows European sibling, DriveNow, and
over the last two decades. The year 2011 stands as the Daimlers car2go electric fleets operate in some cities
beginning of the modern era of lithium-ion battery electric with a high density of charging infrastructure, including
and plug-in hybrid vehicles, with the introduction of the Copenhagen, Vienna, and Amsterdam. In Seattle and
Nissan Leaf and Chevy Volt. Since then there has been Portland, ReachNows fleet is composed of roughly 20%
a steady increase in global sales, with cumulative global BMW i3 EVs, with the balance in other BMW and Mini
sales reaching 1 million in 2015 and recently passing 2 gasoline cars.
million. In 2016, China led the world in EV sales at about
630,000, while Norway led in EV market share of its In contrast to these success stories, in 2016 car2go
countrys total auto sales at 29% (Lutsey, 2017). replaced approximately 400 Smart EVs serving 40,000
members in San Diego with gasoline Smart cars, citing
Although they are not plug-in vehicles, hybrids have a lack of charging stations, range anxiety, and a 20%
proven both cost-effective and popular for taxi fleets. As unavailability rate due to charging time or a low state of
an example, nearly 66% of the New York City medallion charge (Garrick, 2016). This outcome speaks to charging
taxis were hybrids in November 2016 (calculated from density as a prerequisite for EV use in car share fleets.
NYC Open Data). Meanwhile plug-in vehicles have no
significant presence in taxi fleets and mixed success in car Overall the technology and market outlook for EVs
share fleets. appears promising, though the timing of when the
technology will translate into a revolution remains to be
Vehicle Technology Challenges seen. As we discuss in the policy sections of this paper,
it seems likely that strong supporting policies will be
Despite the progress in EV sales, growth has been needed for many years to achieve a full transition from
inhibited by several factors, including limited driving range todays dominant ICE vehicles.
(of most fully-electric vehicles), higher vehicle purchase

Automation vehicle operation. This choice reflects a necessity to
reach complete automation to see the types of major
The technology to enable vehicle automation has impacts on travel patterns that we assume are associated
emerged rapidly, causing much excitement and with automation, such as people willing to spend more
generating much attention, but currently little driving time in vehicles, and vehicles designed with passengers
automation is used in commercially available vehicles. much more than with drivers in mind. In a fully AV world,
Significant technological, legal, and cultural hurdles must especially where these vehicles are privately owned
be addressed before fully automated vehicles take to the and operated, we expect to see some larger, more
roads. This section describes the types of automation comfortable vehicles (perhaps even some with sleeping
available, discusses associated costs, and provides capability), and expect to see some increase in the
insights on when these technologies are anticipated to amount of driving these vehicles (and people who own
move forward. them) do each year. Empty running (i.e. zero-occupant)
vehicles may also emerge as a significant new source of
Automated Technology traffic.
The U.S. Society of Automotive Engineers (SAE) (U.S. Incremental Purchase Cost of Automation
Department of Transportation, 2016) defines a full range
of automation levels (SAE, 2016). Level 1 is widespread One major question surrounding AVs is their cost. There
and level 2 is rapidly being introduced in many models. is a wide range of reported costs of current prototypes
Level 3, including hands-free driving, is just emerging compared to conventional vehicles. Future cost estimates
and only legal in some areas. Levels 4 and 5, with true suggest strong reductions over time, as shown in Figure 1.
full-time driverless operation, is not known to be fully
legal anywhere in the world as of early 2017, except for A key reason for the large spread and rapid reduction in
operation by test fleets. cost estimates by 2020 is the rapid decline in some key
component costs. The cost of LiDAR (Light Detection and
For this study, when characterizing and projecting Ranging) was estimated to be near $75,000 in 2014, and
automation we only consider levels 4 and 5 driverless by early 2016, Velodyne began selling a form of LiDAR for
Figure 1. Estimates of past and projections of future incremental cost of AVs over conventional vehicles. EnoTrans
estimates are low for 2013 because they assume mass production. BCG: (Davies, 2015); EnoTrans: (Fagnant &
Kockelman, 2013); Google Car: (Googles Autonomous Vehicle, 2012); IHS: (IHS Markit, 2014); RMI: (Johnson &
Walker,2016); Tesla: Tesla lists Enhanced Autopilot for $6,000 and Full Self-Driving Capability for an additional
$4,000 on its web site when you configure any of their cars. While cars can be ordered with this functionality, Tesla
has not enabled full self-driving capability on the software end as of this papers publication.

$500 per unit to Ford (McFarland, 2015). Furthermore, the Timing of Autonomous Vehicles
components used for full automation may change over
time, or may vary by manufacturer. For example, Tesla Perhaps the biggest question with AVs is when they will
has indicated no plans to use LiDAR. It seems reasonable really hit the streets. While some analysts still believe
that by 2025, AVs will cost, at most, $10,000 more than it may be decades before these vehicles overcome all
equivalent conventional vehicles. Similarly, electrification technical and legal barriers, many automakers are stating
of the drivetrain and cost of batteries together may they will have models ready in the near future, including
cost about $10,000 more than a conventional vehicle, some indicating a 2020 or 2021 time frame, as shown in
reflecting declining battery costs but rising energy storage Table 1.
of batteries on the average EV. Our EV cost projections
are roughly consistent with reports such as (McKerracher AVs are restricted in most countries at this time, however,
et al., 2016), though somewhat lower than reports that use policies to reduce some restrictions are emerging.
higher future battery cost projections such as (Elgowainy Hands-free driving and testing of fully driverless vehicles
et al., 2016). are allowed in certain areas. In the United States,
the Michigan state legislature recently passed a law
We assume that the two together cost about $18,000 permitting automakers to operate networks of self-driving
more than a comparable ICE vehicle. This drops to about taxis within the state, perhaps one of the first jurisdictions
$10,000 by 2050. Even by 2030, autonomous EVs used in the world to do so. Overall we assume a limited rollout
intensively in car sharing roles are estimated to earn back of driverless vehicles through the early 2020s in leading
the higher purchase cost from energy savings within their countries, followed by mass market rollouts beginning
first three years of operation. about 2025.
Table 1. Examples of AVs in development
Company Vehicle Brand Model Powertrain Production Notes

Goal
Nissan Nissan Leaf Electric 2020
GM Chevrolet Bolt Electric Testing 40 cars in

San Francisco and
Scottsdale
FCA Chrysler Pacifica Hybrid Plug-in Hybrid Testing 100 vehicles

with Google
Ford Ford Fusion Hybrid 2021
Volvo Volvo XC90 Hybrid
Uber Ford Fusion Energi Plug-in Hybrid
Uber Volvo XC90 Hybrid
Daimler Mercedes- F 015 Luxury in Hydrogen Fuel Research Vehicle

Benz Motion Cell Plug-In
Hybrid
Hyundai Hyundai Ioniq Electric Testing 3 vehicles in
South Korea
Hyundai Hyundai Tucson Hydrogen Fuel Testing 2 vehicles in

Cell South Korea

Potential Energy Impacts of Large-scale Vehicle Automation
A wide range of potential impacts of full vehicle automation have been discussed in the literature (Beiker &
Meyer, 2014). In terms of energy use (and consequently CO2), the range of potential impacts is estimated to be
wide and uncertain, due to impacts on many aspects of travel and vehicle efficiency (Brown, Gonder, & Repac,
2014; Wadud, MacKenzie, & Leiby, 2016). As shown in Figure 2, these include improved technical vehicle
efficiency, eco-driving, reduced traffic congestion and platooning. On the other hand, reductions in travel
cost and new traveler groups could lead to significantly more driving, while faster driving and increased use
of energy-using features could lead to more energy use per kilometer. The net effects tend toward significant
increases in driving and efficiency, with a wide range of possible net impacts on energy use, from large
increases to large decreases. As discussed later in the report, we use fairly conservative estimates on most of
these impacts and their combined effects, but acknowledge the uncertainty.
Figure 2. Potential changes in energy consumption due to vehicle automation

Source: Wadud, MacKenzie, & Leiby, 2016
Shared Mobility
Shared mobility has grown substantially around the world in the past five years with the introduction and growth of
many new business models. The following definition is illustrative:
Shared mobility the shared use of a vehicle, bicycle, or other mode is an innovative transportation
strategy that enables users to gain short-term access to transportation modes on an as-needed basis.
The term shared mobility includes various forms of carsharing, bikesharing, ridesharing (carpooling
and vanpooling), and on-demand ride services. It can also include alternative transit services, such as
paratransit, shuttles, and private transit services, called microtransit, which can supplement fixed-route
bus and rail services. With many new options for mobility emerging, so have the smartphone apps that
aggregate these options and optimize routes for travelers (Shaheen, Chan, Bansal, & Cohen, 2015).
The terms ride hailing and ride sharing have become somewhat equated in common usage but should be kept
differentiated, since a hailed ride is not necessarily a shared ride.

Ride Sharing Versus Ride Hailing, and Related Terms
There has become considerable confusion around and misuse of the terms ride sharing and shared
mobility in recent years. We clarify these terms here, and contrast these with ride hailing. We also
introduce the term trip sharing. There are two basic concepts:
Ride sharing (or trip sharing or shared mobility) This refers to rides or trips that are actually shared
between different individuals or different parties and paid separately. It can also more broadly include
public transit services.
Ride hailing (or ride booking) This refers to any app-based system to secure a ride from a taxi or other
on-demand ride service provider such as GrabTaxi, Uber, Lyft, Ola, Easy Taxi or other TNCs. These rides
may or may not be shared.
It is important to keep the two concepts separate. On-demand ride-hailing services are not ride-sharing
services unless they exclusively offer shared rides (such as a micro transit bus system). We consider it
misleading to use the terms ride sharing and shared mobility to refer to a ride-hailing service in a
generalized manner, and while we use these terms to refer to any situation with truly shared trips, we try to
avoid their use for the more general situation of ride hailing, throughout this report. We also use trip sharing
in this report to emphasize true shared mobility, and to avoid overuse of the terms ride sharing and shared
mobility.
The following section discusses the future potential of services is receiving considerable attention in the United
each shared mobility type and explores the ability of States, with a key question being whether rides tend
sharing to decrease the total number of vehicle kilometers to substitute for higher CO2 trips. At issue is whether
traveled on streets by increasing the occupancy of TNCs help cut car ownership and use rates generally,
vehicles for trips. or whether they compete with public transport, thereby
undermining CO2 benefits (Alba, 2015). Another question
Ride Hailing and Ride Sharing is whether shared rides account for a significant share of
trips.
The rise of TNCs like Ola, GrabTaxi, Uber, and Lyft has
been especially rapid since about 2012. Launched in In any case, ride sharing use is spreading to all parts of
2009, Uber reached 1 billion trips worldwide by the end the world, at least to major cities. Figure 3 shows maps
of 2015, and 2 billion within the following six months. from late 2016 for South America, China, and Southeast
Currently, the environmental impact of ride-hailing Asia reflecting the spread in these services.
Figure 3: EasyTaxi availability in South America (left); KuaidiONE availability in China (middle); GrabTaxi availability
in Southeast Asia (right)

Potential Impacts of Largescale Ride- Car Sharing and Bike Sharing
Sharing Services
As of 2017 there is a much smaller market for car sharing
If most people switched to ride-sharing services, with than ride sharing in most countries; however, the general
mostly shared trips, instead of driving their own vehicles, concept of car sharing offers a transportation solution
the numbers of vehicles that would be needed to for users who don't own a car but would like occasional
move people around would decline dramatically both access to a car for more than a single short trip. Car
because of higher load factors and the intensive use sharing offers the benefit of serving more people per
of each vehicle. A recent MIT simulation found that, vehicle than if those people were to use private vehicles,
mathematically and logistically, a fleet of 3,000 vehicles resulting in less need for parking and user cost savings
with capacity of four passengers, or 2,000 vehicles with through more efficient vehicle utilization. Environmental
capacity of ten passengers could meet 98% of the trip benefits can be achieved if the car share vehicles on
demand of 14,000 New York City taxis, with ride wait times average have lower emissions by being of newer model
averaging just 2.7 minutes (Alonso-mora, Samaranayake, year than private vehicles they replace, and by inducing
Wallar, Frazzoli, & Rus, 2017). Similarly, the International a net reduction in VMT because of consumers perceived
Transport Forum developed a simulation of Lisbon higher per-mile costs.
showing that the city could serve its typical daily travel
patterns with only 10% of the vehicles currently used, with Car share business models include traditional round-trip,
a combination of 8- and 16-passenger vehicles (OECD one-way and free-floating, peer-to-peer, and fractional
International Transport Forum, 2015). Of course, these ownership. Within these models, charges can include an
simulations assume people are quite adaptable, ready hourly rate, and in some instances, a per-mile charge as
to make major changes to how they conduct their daily well.
travel, abandon the use of their own vehicles, and are
willing to get into vehicles with strangers. (In a world with Bike share platforms are becoming increasingly popular
small autonomous taxis, rides between two people would for commuters in dense urban areas worldwide. Most
be shared without even a driver on board). systems are station-based and allow one-way trips,
though dockless cycling systems have emerged, with
While ride sharing can cut the numbers of vehicles in some advantages (Handy, 2017). Electric bicycles, or
use, cutting traffic is not assured: for example a system e-bikes, can offer an even more utilitarian transportation
dominated by single-occupant ride-hailed vehicles would option than regular bicycles in bike share. In a recent
not reduce vehicle trips compared to similar private study, e-bikes were found to be used for trips twice the
vehicle trips; moreover, rides could be shifted from public distance of regular non-electric bike share bicycles in
and/or active transport. The International Transport Forum China (Campbell, Cherry, Ryerson, & Yang, 2016). In our
(ITF) study showed that a shift from walking and cycling to previous High Shift Cycling study we found that bike
shared vehicle travel could result in a significant increase sharing has reached more than a thousand cities in
in vehicle kilometers traveled, thus increasing congestion the past few years, though still represents a tiny share
and travel times, even as vehicle occupancies increase of cycling trips around the world (Mason et al., 2015).
(OECD International Transport Forum, 2015). A recent Continued growth could change this, and bike share
analysis of New York City indicates that the rapid increase systems can also introduce many new people to urban
in ridership in TNC vehicles in 2015 and 2016 coincided cycling, who eventually acquire their own bicycle.
with a decline in bus and metro travel during this period
(Schaller, 2017). A world of ride hailing, but without true Dynamic Shuttle Services
ride sharing and strong public transport and active travel
aspects, could be a very high traffic, congested, higher A dynamic shuttle is a smaller shuttle bus that can serve
energy-use and CO2 world. more passengers than a taxi, and offers a more flexible
transportation solution than traditional fixed-route public
Thus the success in ride sharing as an energy-efficient transport buses. Larger on-demand bus systems also
and space-efficient mode will depend both on the exist, such as Chariot and Bridj, with the prospect of
average number of riders per trip, which must be widespread on-demand microbus services around the
significantly higher than modes like private automobiles, world as a potentially optimal size and low-cost travel
and on its ability to draw riders from these less-efficient option. This approach can also help public transport
modes, rather than from public transport services. This is agencies become more efficient and cut costs, by
particularly important in countries where public transport providing targeted on-demand services in areas that have
ridership and vehicle load factors are very high i.e. most trouble supporting standard fixed-route services due to
of the developing world. insufficient ridership.

However, in the event of an increase in the use of on-
demand services, it will be important that systems
and governments ensure that shared mobility remains
affordable for a range of people who depend on public
transport, including people with disabilities, older adults,
and low-income passengers who have benefited from
subsidized public transit services (Polzin, 2017).
Increased use of shared mobility systems, public transport

and active travel (walking and cycling) may present the
greatest potential of the three revolutions to usher in an
era of sustainable transportation in cities. The benefits
they offer include traffic reduction, energy and emissions
reductions, and lower overall systems costs (Handy, 2017).
We explore this in the scenarios presentation that follows.

4. Future Scenarios: BAU, 2R, 3R
Given the major interactions associated with the three transportation revolutions, as well as for economy of
presentation, it makes sense to consider our 3 Revolutions scenarios together, and in the context of the business-as-
usual case. Table 2 presents key characteristics included in each scenario. As can be seen, the scenarios are built as a
series of progressions adding an additional layer at each step. The basic characteristics of each scenario follows.
Table 2. Key characteristics of the 3 Revolutions scenarios
Use of Use of Use of Shared Urban Aligned with

Automation Electrification Vehicles Planning/ 2C (or Lower)
Pricing/TDM Scenario
Policies
BAU, limited Low Low Low Low No

Intervention
2R with high High High Low Low Maybe

electrification,
automation
3R with High High High High Yes

high shared
mobility,
public
transport,
walking and
cycling
The BAU Scenario

As presented in the previous reports, the projected future people in poorer countries wont have access to private
growth of urban travel around the world is several fold, cars in 2050). As a result of this city growth and income
and up to tenfold for particular countries and regions. growth, mobility levels will skyrocket. For example, the
Total urban (metro area) population is projected by the IEA projects a nearly tenfold increase in car travel in India
United Nations to increase by 60% from about 4 billion between 2010 and 2050.
people in 2015 to 6.5 billion in 2050; these urbanites
are projected to collectively become more than twice as Our BAU scenario reflects these projections. While
wealthy as the average urban dweller worldwide is today many cars will exist outside urban areas, just the urban
(with poorer countries such as India seeing a fivefold or population of cars around the world grows from 750
greater increase in incomes, though to levels that remain million in 2015 to 2.1 billion in 2050. Traffic congestion
far below OECD countries, and many or even most increases commensurately even if many new roads are

built to accommodate traffic increases, countries will have regions (the United States, Europe, China and Japan).
a very hard time keeping up. Cities around the world are Other nations follow and by 2040 automated EVs
already heavily congested with car traffic, sometimes with dominate LDV sales worldwide. EVs also dominate
only a very small level of car ownership. Whether the BAU sales of 2-wheelers and buses after 2030 worldwide.
projection of cars and car travel can be accommodated
without complete seizures of road traffic networks is itself By 2050 few non-EVs are sold anywhere. Some
a fair question, and thus our BAU scenario that assumes EVs may well be plug-in hybrids, but even these we
functional road systems may be unrealistic in this regard. assume are basically phased out by 2050 as longer
range EVs and fast charging become ubiquitous.
This future also assumes that investments into alternative
transportation modes everything from buses and rail By 2030, EVs will have an average range of 250
systems, to bicycles and even an extensive system of safe miles (400 kilometers) and an incremental cost
sidewalks in cities lags behind what is needed to retain of about $10,000 per vehicle. Costs could decline
current mode shares, as more people gain access to cars more than this, but our assumed increase in driving
(as well as motorized 2-wheelers). This is a world where range requires larger battery packs, which we take
people fear crossing the street due to traffic, fear cycling into account in our cost estimates. These incremental
due to lack of safe cycling infrastructure, and sometimes costs continue to decline to near zero in 2050. EVs
fear taking public transport due to personal safety in 2030 save enough on energy costs to pay back
concerns. This is also a world where public transport within five years, even less for high-distance drivers.
systems are not well designed and do not have adequate
investment to ensure they are of high quality and high By 2050 electricity grids are substantially
capacity. We do not see rapid growth, for example, in bus decarbonized worldwide. In our main 2R scenario,
rapid transit systems, which allow buses to cut through we assume that electricity grids are decarbonized
gridlocked car traffic and move people faster than they at a steady rate, consistent with the IEA 4DS (4C
can in private vehicles. scenario). This means that electric AVs can achieve
significant CO2 reductions, even if they drive many
This future results not only in possible gridlock in the trillion kilometers per year. However, as we show
worlds cities, it likely fails to stem the trend of high injury below, in a 2C or lower world, grids must be nearly
and death rates on the worlds roads over 1.2 million completely decarbonized worldwide by 2050.
deaths in 2015 alone. And a world that continues to We consider this as a sensitivity case where EVs
rely on ICE vehicles, even though these will continue to worldwide become truly zero-carbon vehicles by
become cleaner, may have trouble achieving truly clean 2050.
air in its larger cities. This is on top of our BAU projection
of a 50% increase in energy use and CO2 emissions A critical assumption is supportive government
during a period where CO2 emissions must drop policies. This scenario probably does not happen
dramatically to achieve a 2C or lower temperature limit to without strong policies to encourage uptake of
arrest climate change. EVs as they continue to mature. A range of policies
are already in place in many countries (including
incentives for installing charging infrastructure,
The 2R Scenario: Electrification access and parking advantages, and tax incentives)
and these must continue as EV markets develop. In
and Automation particular, it seems likely that vehicle price incentives
will be needed through 2025 or 2030 given ongoing
This scenario includes both a rapid increase in electric incremental first costs, though what levels will be
(non-autonomous) vehicles and, later, autonomous EVs. needed over this time frame to spark and sustain an
There are many important assumptions and these are laid S-curve revolution are far from clear.
out below.
Electrification rates of new LDV sales in two example
Assumptions for EVs countries, the United States and India, are shown in Figure
4. Private vehicles experience slower electrification than
By 2020, 5 million EVs are sold annually worldwide commercial vehicles until about 2030 then catch up.
(compared to 750,000 in 2016), with sales
continuing to rise sharply thereafter. The steep part
of the S-curve of sales for EVs occurs between
2020 and 2030 in the worlds leading nations and

Figure 4. EV sales shares for selected countries, vehicle type, years and scenarios
Assumptions for Automated EVs

Commercial availability of fully driverless vehicles after 2025. By 2030 sales are entering a steep
begins shortly after 2020. In the more advanced S-curve phase, where most commercial enterprises
economies such as the United States, Europe, Japan go driverless, and where most households in leading
and China, vehicle automation (and policies that countries choose to buy a driverless car by 2035.
allow or promote it) advance in the coming five years Other countries follow by five or at most 10 years, with
through various levels, such as hands-free driving, a high level of driverless vehicle sales in all countries
fully autonomous capability but with a driver in the by 2045.
seat, to fully driverless (SAE level 4: no steering
wheel or pedals) by 2020 or soon after (SAE, 2016). AV sales start with commercial operations, but
Such vehicles enter full commercial operation with households follow soon after. It appears likely
largescale production by 2025. that the strongest business cases for adopting
driverless vehicles will be commercial operators who
The cost of driverless vehicles declines rapidly, as otherwise pay their drivers, so it can be expected
it has already started doing. While in small volumes that these (including TNCs and some public transport
these vehicles may costs tens of thousands of dollars operations) will be early adopters. However, it also
more than conventional vehicles, we assume that by seems likely that pioneer households will adopt
2030, in reasonably large volumes these vehicles are these vehicles as soon as possible. Thus we have
about $10,000 more expensive than conventional them follow soon after, and as part of the 2R scenario,
vehicles, and down to $5,000 more by 2040, apart these households use their own AVs into the future
from possible new features such as entertainment rather than significantly increase use of shared
systems in more comfortable, larger interiors. The mobility systems.
combination of electrification and automation is
estimated to cost close to $20,000 more than
conventional vehicles in 2030, declining to below
$10,000 more in 2040. Energy savings over the life
of the vehicle offset some of the higher first costs,
especially for commercial, high travel, vehicles.
Due to policies described below, fully driverless
vehicle sales in leading countries ramp up rapidly

As shown in Figure 5, AV sales are assumed to increase rapidly after 2025 in the 2R scenario and reach up to 25%
market share in 2030 in leading countries. Commercial vehicle fleets are assumed to be rapid adopters and reach
higher shares by 2030, but since there are many more private vehicles in the 2R scenario, the total numbers of AVs in
households eventually far surpasses those in commercial fleets. In trailing countries such as India the process happens
more slowly, but sales of AVs to commercial operations still approach 20% by 2030, with households and fleets
reaching 50% or more by 2050.
Figure 5. AV sales shares for selected countries, vehicle types, years, and scenarios
Household-owned driverless cars will be larger and in contrast to privately owned vehicles, we assume
more comfortable. AVs may be designed to be much only a small travel increase for publicly shared AVs
more comfortable and to better support non-driving since their use will be paid per kilometer of travel.)
activities, with amenities such as mobile offices
or with home theatre video systems. We assume a The rapid rise in the use of driverless cars in
significant increase in SUV-sized vehicles (or even households precludes a rapid growth in shared
larger vehicles such as vans but with similar weight mobility. People are content to continue to travel in
and fuel economy as todays larger SUVs), offsetting their own vehicles, which are now more comfortable
some of the energy savings of electrification. And with and can be sent on errands without occupants.
the increased comfort and elimination of driving, the This reinforces the ownership model that is already
time cost of driving in AVs will be significantly lower, attractive and leads to high car ownership rates
since people wont mind being in their vehicles for worldwide by 2050, similar to the BAU. Ride hailing
longer periods of time when they do not have to drive serves a niche activity in cities as it does today.
them, and can conduct other activities.
During the transitional (roughly 2025 to 2050)
Given the reduction in time cost, people will drive time frame there remain many legacy vehicles that
in their vehicles significantly more than they do require drivers. We estimate that even with a 10- to
today. In all regions of the world, we assume a 10-15% 15-year transition in sales in the leading countries,
increase in driving per capita (and per vehicle) in there would remain significant non-AVs on roads in
personal AVs relative to BAU. This could also include 2040 and still some remaining by 2050. The mixture
increases in zero-occupant vehicle travel, as people of the two types of vehicles will create its own issues
assign vehicles to conduct tasks such as retrieving and perhaps hamper some of the efficiency and
family members or even packages. We assume congestion reduction benefits of the driverless cars.
another 5% increase in vehicle travel from this in our We do not attempt a detailed analysis of this issue but
scenarios, resulting in an overall 15-20% increase in flag it for further study.
vehicle travel, though we acknowledge the effect
could be more significant. (As described for 3R below,

The 3R Scenario: Adding Another feature of the 3R scenario is a revolution
in types and roles of public transport, along with a
Shared Mobility steady increase in its use around the world. After
2020, public transport services become more
The third scenario overlays shared mobility and strong tailored to a shared vehicle world. Major travel
policies for urban planning that favor compact cities, corridors continue to be served by efficient bus
walking, cycling, and public transport. In the 2R scenario, systems such as bus rapid transit, and major cities
solo car travel becomes more convenient and cheaper continue to build rail systems for the busiest travel
per mile. Public transport systems also get cheaper, with routes. Smaller buses, with 8-16 seats, grow in
drivers eventually eliminated, but still have a difficult time number, as these are almost capable of providing
competing with private cars and 2-wheelers. In the 3R point-to-point services and can be summoned at
scenario, we assume the opposite is true: sharing rides least to locations nearby specific residences if not
(in the form of on-demand ride hailing services as well as to the door. Even with a driver, on-demand small bus
public transport) becomes extremely popular. Some of this and van services provide a very low-cost, convenient
may occur just from a cultural change in countries around travel option for many types of intermediate trips in
the world, though as we describe in the policy section dense areas. As vehicles become automated, the
below, we dont expect this type of scenario to occur cost of small-bus travel drops further to become the
without strong policies to give shared mobility, especially cheapest per-passenger-kilometer on-demand travel
high occupancy public transport, cycling, and walking, a option in the world.
distinct advantage over private, single-occupant travel.
The result of all these changes in 3R is an
Assumptions for the 3R Scenario ecosystem of public transport and ride hailing
services that are harmonious and complementary.
By 2020 shared mobility represents a significant Small vehicle ride hailing does not displace trips
share of urban travel in most major cities of the from larger public transport services, except where
world. Thus shared mobility gains substantial traction currently large vehicle public transport is poorly
before automation even begins. This includes a utilized and inappropriate given corridor demand.
range of ride-hailing services and vehicle types, with One result of this ecosystem is significantly higher
more right-sized vehicles for different types of trips. load factors (average passengers per trip) in all
Average load factors (people per trip) rise significantly vehicular modes.
in countries (such as the United States) where it is
currently below two, and stays high in countries (such This scenario also features a range of policy and
as India) where it is already high. In 2015, most of the planning initiatives to make cities much more
world experienced well over two people per ride in cycling and pedestrian friendly. Sidewalks and bike
LDVs. This is preserved and even increased in the 3R lanes are added to create continuous networks and
scenario. ensure maximum safety for these travelers. A general
effort to develop more compact cities with shorter
Car sharing also grows, but since automation begins trip requirements is also assumed, with trip lengths
to increase rapidly after 2020, we assume that by dropping by 10-15% compared to the BAU scenario,
2030 car sharing is indistinct from ride sharing in rather than increasing as they do in the 2R scenario.
both cases an empty vehicle is hailed on the street These efforts are crucial to preventing new vehicular
or summoned from a parking place to provide travel options from displacing large numbers of
mobility services. We also assume that taxis become walking and cycling trips by lower costs and inducing
indistinct from TNCs all ride-sharing services use more dispersed development patterns which increase
apps and encourage sharing via pricing systems. travel distances and make cycling and walking
infeasible.
One result of this revolution in on-demand mobility
is a steady decline in privately owned vehicles. After Figure 6 shows the sales of private and shared LDVs in
2020 sales and use of commercial TNC vehicles the United States and India by year for the BAU and 3R
rises rapidly and by 2025 there is a resulting decline scenarios. It also breaks out these sales bars by ICE, EV,
in purchases of household vehicles, although legacy and automated EV to give a sense of how these three
stocks of household cars remain for over two decades revolutions interact in the 3R scenario. Compared to the
and could create a glut of unneeded private vehicles. BAU as well as 2R scenario, the 3R scenario reaches
We assume these cars are driven less and less over high levels of light-duty shared mobility vehicle sales by
time. 2030, with India (and all other countries) not far behind

U.S. levels (between 40% and 50% sales by 2030). In the United States, most of these are either electric or automated
and electric, whereas in India most are ICE and not automated. However, by 2050, nearly all new TNC vehicles are
automated and electric, and account for over 75% of LDV sales worldwide.
Figure 6a-b. Sales of private and shared vehicles for United States and India by scenario, year and vehicle type

True Shared Mobility?
One major question across these types of shared mobility, in particular for ride sharing, is whether many of
these trips are truly shared i.e. by separate travelers making these services really different from classic
taxi services. The extent to which rides are shared in services in different parts of the world is not well
documented. A typical car trip in the United States has fewer than 1.3 people, while in a country like India it
might have well over two.
Many TNCs offer incentives for ride sharing, passing through the savings inherent in adding people to the
ride. Lyft has reported that up to 50% of their riders have opted for their lower-cost carpool Lyft Line platform
in cities where that option is available. If this type of sharing occurred at much higher volumes in the future
and was not the result of fewer people walking, cycling, or taking public transport, the impacts on travel
would be profound. Far fewer vehicles would be needed to move a given number of travelers.
But it is not clear this will occur, at least without policy support to encourage it. For example, one dynamic
between shared mobility and automation is that driverless ride-hailing services may become so inexpensive
that the incentive to share rides to save cost will be substantially reduced. Even if most vehicle trips are
shared, if a significant portion of those trips were formerly made by walking, cycling, or public transport, the
amount of vehicle travel could increase significantly, increasing congestion and reducing the ability of many
people to access opportunities.
In designing our 3R scenario, we assume that a range of policies are implemented that promote true ride
sharing and result in 30-40% more passengers per vehicle in 2050 than in the other scenarios. (In the United
States, for example, this would represent an increase from 1.3 to 1.8 passengers per car trip.)
Results: Passenger Travel Projections Across the Scenarios

By combining all of the assumptions and partial projections shown in the previous section with our scenario tool, we
can create a complete picture of passenger travel by mode worldwide, summing across the eight countries and regions
in this study (Figure 7). These results reflect myriad assumptions and estimates, but also some basic arithmetic: Total
pkm by mode is equal to the number of vehicles by mode, the annual travel of these vehicles, and their average load
factor.
Figure 7. Total passenger kilometers of travel by mode, scenario, and year worldwide

For example, the much higher travel levels of people comparison to both the 2R and BAU scenarios). This is the
in shared AVs in the 3R scenario in 2050 reflects huge numerical manifestation of the ecosystem of harmonious
increases in the numbers of these vehicles, their higher and complementary travel modes mentioned above. By
travel per vehicle than private vehicles, and their relatively 2050 the vast majority of urban LDV travel takes place in
high load factors. These averages mask many regional on-demand shared mobility services, with private vehicle
differences but are useful to show the stark differences in travel declining steadily over the two-decade period as
travel across the scenarios. private cars eventually are scrapped and not replaced.
On closer examination, a number of notable results The impacts on vehicle travel, taking into account the
are shown in Figure 7. In 2030, across all eight regions numbers of vehicles and their use to fulfil the projected
studied, there are still relatively few AVs plying the worlds passenger travel by mode, is shown in Figure 8. Here
roads. This reflects the fact that only leading countries the differences between 3R and the other scenarios is
are deeply into selling these vehicles in that year, and greater, since travel in the 3R scenario is supported with
the vehicles still represent a small share of stocks even fewer vehicles and vehicle kilometers carrying more
in these countries. However, by 2050 in the 2R scenario, people per trip. In fact the growth in vehicle kilometers
well more than half of private vehicles are driverless. This between 2015 and 2050 worldwide in the 3R scenario is
has the effect of increasing driving rates significantly in 2R only about 30%, even though passenger travel grows by
compared to both the BAU and 3R scenarios. Assuming over 60%. In contrast, vehicle travel grows dramatically in
the time cost of travel is halved, along with the possibility the BAU scenario and even more so in the 2R scenario,
of zero-occupant trips, we assume a 15-20% increase in with nearly a tripling between 2015 and 2050 worldwide.
driving per automated car vs. non-automated car, around The 2R scenario also reflects the rapid rise in automated
the world. EV driving after 2030. (Given the already complex nature
of the figure, non-automated EVs are not shown).
In the 3R scenario, significant shared travel does occur
by 2030, along with increased travel by minibus and
other forms of public transport, cycling and walking (in
Figure 8. Vehicle kilometers of travel by motorized mode, scenario, and year worldwide
(walking, cycling and e-bikes not shown)

Considering Rapid Transitions
The scenarios in this report depict very rapid transitions to a different future. How do such rapid transitions
occur? Clearly the new technologies and travel services must be compelling for consumers. The tipping
point concept is important here trends start slowly, technologies improve, more people learn about
them, and at some point sales expand beyond pioneers and early adopters to mass market. We depict
about a 10-year period when the market share of AVs moves from about 10-90%. This certainly reflects
a rapid increase in mass awareness of and desire to own such vehicles. Is this likely? Perhaps not, but if
these vehicles offer enough advantages at an acceptable price, this scenario seems quite possible.
Another question is whether the suite of electrification and automation technologies can evolve this
quickly and costs can come down by, lets say 2025, to support this revolution. It also depends on enabling
policies. Finally, it will depend on the ability of vehicle manufacturers to shift plants and equipment, an
investment challenge on the order of trillions of dollars. The faster this must happen, the more expensive it
could be from the point of view of both retiring useful equipment and raising capital.
Vehicle Sales and Stocks

As shown for electrification, automation and shared mobility in a previous section, the sales increases and changes
in mode shares assumed in these scenarios are rapid and can have fairly profound effects on vehicle stocks. In fact,
personal and business decisions around vehicle holdings are what determine sales in the first place. Using an annual
stock adjustment model and typical scrappage rates for LDVs, we estimated the impacts on stocks from the LDV sales
trends in our scenarios. In Figure 9 we show the sales and stock effects for the 2R and 3R scenarios in the United
States, as an example.
Figure 9a-d. LDV sales and stock evolution in the 2R and 3R scenarios, U.S. example

There are at least two important trends: in TNC applications are used much more intensively,
with higher passenger load factors. There is also
The first is that in the 2R and 3R case, stocks of lower overall travel demand and higher mode share
conventional vehicles last far beyond their declining for non-LDV modes in this scenario (not shown in
sales shares. This is not surprising. But it does raise these figures). The combined effect is that by 2050,
a question: if a conventional vehicle bought in, say, urban LDV stocks decline by about 70%, from about
2025 lasts until 2045, how does the world rapidly 2.1 billion in the BAU and 2R scenarios to about 500
transition to electric, automated or shared vehicle million in the 3R scenario. This takes us back to the
systems? The answer is, its difficult, and this legacy sales figures, which also decline substantially (by
stock of conventional vehicles could hamper such about 40% in the U.S. example) in the 3R scenario,
a transition. We found that in order to assume these since far fewer LDVs are needed to meet the travel
vehicles dont somehow result in additional travel that demand. The reduction in sales is not as significant as
isnt otherwise needed, many would either have to the drop in stocks because each vehicle drives about
be scrapped early or simply not driven, while other five times more per year than a private vehicle and
vehicles are used instead. A detailed analysis of such is turned over in four to six years instead of 20 to 25
dynamics is beyond the scope of this project but is years. Sales are adjusted to meet this higher turnover
an interesting area for further study. We do assume rate, but the net effect is lower global sales of cars
a slow but steady decline in the use of existing on a worldwide basis, 2050 urban sales in the 3R
conventional vehicles in the 3R scenario between scenario are around 60 million less than half of the
2030 and 2050, with about a 25% reduction in 135 million reached in the BAU and 2R scenarios and
annual per-vehicle travel by 2050 compared to those effectively a return to current levels (which are about
vehicles in a BAU scenario. two-thirds of the 90 million, urban + non-urban, sold
worldwide).
A second is that in the 3R scenario, total stocks of
vehicles decline dramatically since shared vehicles
Figure 10a-d. Evolution of vehicle and pkm of travel, U.S. example

Light-duty Vehicle and Passenger Travel United States, ride sharing is starting from relatively low
averages and will need to rise in the 3R scenario, whereas
The changes in LDV sales and stocks are aligned with in India it is already high and the trick will be to preserve
the passenger demand for LDV travel in each region and this high level of sharing.
scenario. Looking at vehicle kilometers of travel and pkm
in those vehicles reveals a few more dynamics. In the 3R scenario many more shared trips may be taken
with two, three, or more passengers. Some such trips
Results in terms of vehicle and passenger travel over time occur in the other scenarios as well, naturally. But a net
are shown in Figure 10 for the 2R and 3R scenarios in increase of close to 0.5 passengers per vehicle in the
the U.S. case. Notable is the much higher vehicle travel United States would be fairly dramatic. For example, it
level in the 2R scenario, especially after 2030. This is also would occur when shifting from 80% single-, 10% double-,
higher than in the BAU case, about 15% higher by 2050. and 10% triple-occupant vehicle shares to a 50% single-,
This reflects our assumption of the travel rebound effect 20% double-, and 30% triple-occupant share of these load
due to lower out-of-pocket costs as well as much lower factors. Such shifts could be driven by strong marketing
time costs of travel in AVs. As expected, vehicle travel by and pricing strategies on the part of TNCs to encourage
shared vehicles in the 3R scenario is far higher than their shared trips, as well as by policies to promote this sharing.
stock share, given the intensity of use of these vehicles. See further discussion in Section 5, Policy Narratives.
These supply a high share of overall LDV travel in the 3R
scenario by 2050. By that year, total LDV travel in the 3R Average load factors also rise in the 3R scenario for public
scenario is about half that of the 2R scenario. transport services, since they are better coordinated with
ride sharing services, and we assume people in denser
Finally, passenger travel is simply a reflection of vehicle cities are more amenable to taking transit. In addition,
travel, with passengers per vehicle factored in. We do not with widespread TNC light-duty vehicle services, some of
assume any significant difference in passenger loadings the lowest productivity public transport routes could be
between the BAU and 2R scenarios. However, in the 3R eliminated, particularly in countries like the United States
scenario shared mobility results in significant increases where average bus load factors are among the lowest in
in passengers per vehicle, rising over time, as reflected the world.
in the U.S. and India examples shown in Figure 11. This
comparison reveals two very different situations. In the
Figure 11. Average passengers per LDV, U.S. and India examples
Scenario Impacts on Energy and CO2 Emissions

Having established the passenger and vehicle travel Vehicles of all types may become more efficient
shares by region in the various scenarios, it is not difficult over time as technologies improve. We follow the
to estimate the types and amounts of energy used, and IEA BAU projections of fuel economy improvement
CO2 emitted, in relation. In addition to the assumptions given current and expected efficiency improvement
described above for each scenario, there are a number programs and regulations around the world. By 2030
of key features of these scenarios that impact energy use conventional vehicles become roughly 30% more
and CO2: efficient than in 2015, as a stock on-road average.

We assume significant changes in the mix of will be strongly decarbonized over the next 35 years.
vehicles in each scenario that affect efficiency. The IEA projects in its 4C scenario (roughly a baseline
For example, we assume that automation and the or BAU type of scenario) that the average carbon
resulting interest in spending more time in vehicles intensity of electricity worldwide will decrease by
leads to larger vehicle sizes. Vehicles may be about 50% in 2050 relative to 2015. In a 2C scenario,
redesigned, for example, to make it possible for four the average carbon intensity of electricity worldwide
people to sit comfortably facing each other, or to will decrease by more than 95% in 2050; in a 2C
observe entertainment systems, all of which could world, nearly all electricity worldwide is generated
require larger vehicles. We assume that in the 2R from zero-carbon sources.
scenario, larger vehicles offset some of the benefit
of electrification and automation. In the 3R scenario The results in terms of energy use are shown in Figures
we assume far greater right-sizing of vehicles, with a 12 and 13. In Figure 12, LDVs dominate energy use in all
distribution from 2-seaters to minibuses. Light-duty scenarios, although energy use declines significantly from
vehicles are smaller on average in 3R than in the BAU the BAU to 2R to 3R cases, particularly in 2050. In the 2R
or 2R scenarios scenario, the strong uptake of EVs cuts energy use by
about 40% compared to the BAU scenario in 2050, with
EVs also have a singular impact on the use of gasoline some reductions already by 2030, thanks to the rapidly
and diesel, and on other liquid fuels such as biofuels. rising use of EVs at that point. The 3R scenario delivers
The rising use of EVs in the 2R and 3R scenarios, an additional significant reduction in energy use by 2030,
across all types of vehicles, drives oil use to very and a very large reduction even compared to the 2R
low levels by 2050. In fact, from an energy point scenario in 2050 (more than 50% lower), with energy
of view, the main advantage of 3R over 2R is that use less than one-third of the BAU case. This reflects the
the electricity demand in 2050 the electric power combination of electrification and a strong shift toward
needed to support transportation is far lower. shared mobility, higher load factors, and much lower
energy use per pkm of travel service provided around the
Finally, the shift to electric power can help strongly world.
decarbonize passenger vehicle travel, but only if
electricity itself is decarbonized. Countries such as These differences are also reflected in Figure 13, breaking
India currently do not have a power grid mix that energy use into ICE vs. EV consumption rather than by
would offer much, if any, CO2 reduction from a 2R mode. The dominance of electricity in the 2R and 3R
scenario over a BAU scenario. Of course, electric scenarios by 2050 is evident, and since EVs are much
grid mixes will evolve, and given Paris Agreement more efficient than ICE vehicles, the overall energy use is
commitments, there is a reasonable chance that these far lower than in the BAU scenario.
Figure 12. World energy use across all modes, by scenario and year

Figure 13. World energy use, ICE vs. EV, by scenario and year
Breaking Down the 2R and 3R Scenarios Net Impacts on Energy Use
Many factors in our model affect energy use, either by affecting vehicle travel (and the share of travel by
different vehicle types), or the energy efficiency of that travel. Table 3 provides some key indicators, each of
which is affected by different factors.
Table 3. Summary of key travel and efficiency assumptions across scenarios
Factor 2R Assumption 3R Assumption Notes
Private AV travel per 15-20% higher per vehicle Same private AV increase This includes empty
vehicle than BAU in all years as 2R (but with far lower running of vehicles
private AV travel share)
Public (shared) AV travel Similar to non- Same as 2R No "induced travel" effect
per vehicle autonomous shared since travelers pay at the
vehicles (intensive travel margin for each trip
given service provision)
Non-autonomous EV Roughly 50% better than Same as 2R ICE vehicles improve
efficiency similar ICE vehicles over time, with increasing
shares of hybridized and
lighter vehicles
AV/EV efficiency 60% better (lower energy/ Same as 2R Autonomy provides
kilometer) than similar additional efficiency
conventional vehicle benefits, mostly in-use,
despite possibly higher
traffic levels
LDV shared mobility Little effect in this Up to 33% lower LDV Given up to 50% higher
scenario energy per pkm from load factors
sharing
Public transport Major improvements Same as 2R except A high share of public
from electrification and steady or increasing load transport vehicles are
automation though factors result in more automated in most
smaller decrease in energy production per regions by 2050, similar
energy per pkm due to pkm to LDVs
declining load factors

The CO2 emissions associated with these scenarios are shown in Figure 14. Generally speaking, the ICE vehicles use
primarily gasoline or diesel fuel and continue to do so into the future; these fuels remain carbon intensive since no
major increase in biofuels is assumed in these scenarios. Thus, CO2 emissions from ICE vehicles closely track their
energy use. On the other hand, EVs benefit from steady, ongoing reductions in the CO2 of electricity generation. Our
base electricity scenario (shown here) uses the IEA 4 C scenario projection of electricity carbon intensity, with an
average of 40% reduction between 2015 and 2050 worldwide. However, for a 2C climate target to be achieved, as
projected by IEA, electricity-related CO2 emissions must be nearly completely eliminated by 2050. Achieving that would
be reflected in this figure as removing the electricity CO2 (yellow bars) from the chart, leaving only the blue bars. In such
a case the 3R scenario nearly achieves a zero carbon urban transportation energy system in 2050. Either way, the 3R
scenario cuts CO2 more than the 2R scenario, even by 2030 and onward through 2050, due to the greater reduction in
petroleum energy use it provides via shared vehicle trips and greater public transport and non-motorized travel.
Figure 14. World CO2 emissions by scenario and year, BAU EV decarbonization cases
The Paris Agreement and Very Low Carbon Scenarios
The climate targets the world adopted in Paris in 2015 (the Paris Agreement), include well-below 2C or 1.5C
limits to global temperature changes. To achieve 2C, a 50% reduction in CO2 by 2050 relative to 1990 levels
as an average across all energy sectors is often cited as an appropriate global target (IEA, 2016). There
are few published analyses on achieving a 1.5C target but it is not unreasonable to infer that emissions in
all sectors must be very low by 2050, headed toward zero not long after. The exact role for transportation
(and in this case, urban transportation) is uncertain, but clearly it must achieve very low CO2 emissions by
2050 to do its part to achieve the broader targets. A reasonable approximation would be more than a 50%
reduction in CO2 emissions in 2050 compared to 2015 for a 2C scenario, and something closer to zero CO2
emissions for a 1.5C scenario. In the scenarios shown in Figure 14, the 2R case appears on track to achieve
a 2C target, whereas the 3R case is more consistent with a 1.5C target. If electricity generation CO2 drops
to zero worldwide, both scenarios may be 1.5C scenarios.

one kilometer of highway, one square kilometer of
The Costs of 3 Revolutions parking, or one bus, or running a bus system per pkm of
service) are retained from our previous High Shift studies.
Transportation is expensive. There are costs associated Three new estimates are included here:
with constructing and operating vehicles; extracting and
converting energy to its final form as fuel; building and The cost of EVs and AVs, as noted in the assumptions
maintaining roadways and vehicle fueling infrastructure; for the 2R scenario. These costs decline over time but
operating public transport and other systems; and paid remain fairly expensive through 2030.
drivers.
The cost of paid drivers. Vehicle drivers, including for
In our previous High Shift studies (and in IEAs Energy taxis, TNC vehicles, buses, and trains, can account for
Technology Perspectives series), most of these costs up to 50% of the cost of operating these enterprises.
were estimated and projected out to 2050 for countries We did include this cost in previous studies for public
around the world. Here we reorganize this cost analysis transport systems, but taxi and TNC services were
around our 2R and 3R scenarios, and add some important neglected. We have estimated the average cost
elements. We continue to use average costs (as final per driver plus overhead for OECD and non-OECD
prices) for all cost components within each of our regional countries taking into account typical wage rates. This
breakouts, and adjust prices into the future. is naturally an important element in lowering the cost
of operating AVs.
Beyond the cost of purchasing and fueling vehicles,
one of the most important costs, it turns out, is the labor The cost of operating and managing TNC systems.
associated with paid drivers. Nearly as important, at least This is difficult to estimate, given the fairly early stage
in the 3R scenario, is the cost of running companies that of operating such systems, but it appears potentially
manage shared mobility networks (the TNCs of the world). very important in a world with high levels of use of
Finally, the additional cost of EVs and automated EVs over such vehicles. The current 20-25% overhead rate
conventional ICE vehicles is not minor, though this is offset (fee charged per unit revenue charged by drivers)
over time by fuel cost savings. that is typical of systems today is assumed to decline
somewhat as these systems expand, but is kept close
Specifically, the costs in our estimates include: to 20% from 2020 to 2050.
Purchase cost, maintenance cost, energy cost of Finally, the method for estimating and tracking costs
private vehicles (everything from bicycles to large here is based on societal costs. We look at the fuel use
SUVs) over the entire 20-year operating span of a vehicle,
All costs associated with operating TNC services and allocate its capital cost over that same span. We
All costs associated with operating public transport do not discount costs into the future, though we use
services real rather than nominal cost estimates, accounting for
Costs of constructing and maintaining roads, inflation. Of course, private actors make decisions very
sidewalks, bike paths, and parking facilities differently, using short-term considerations, rapid payback
requirements and such. Our cost comparisons do not
We have not attempted to quantify the costs of: provide such context, and thus costs that may be lower
from a long-term, societal point of view, may not appear
Traffic congestion and the associated time delays lower to these agents. Thus policies may (among other
Traffic safety, in the form of crashes, injuries, and things) need to help better align private costs with societal
deaths costs in order to achieve the scenarios we present here.
Air pollution and its impacts on health
Noise and its impacts on health A full set of costs across cost categories and comparing
CO2 and its impacts on the climate though we across modes, for two example countries (the United
do make some cost estimates associated with this States and India) in 2030 is shown in Figure 15. These
pollutant, in the form of cost-per-ton of CO2 reduced. are estimates for BAU scenarios, and the per unit costs
per pkm do change somewhat for other scenarios, but
However, these are all important forms of societal cost generally not by a lot. These figures show some of the
that are affected by the different scenarios, and we main differences across modes and situations worthy of
consider some of the possible impacts below. note.
Most of the cost assumptions per unit (such as building Modes in OECD countries with paid drivers are nearly

double the cost per pkm of modes without drivers. Clearly, companies. This is not surprising since this is a very
removing the driver can increase the competitiveness personalized service, with higher costs all around
of any commercial mode dramatically. This is far more given the small number of people included in each
important than the increased cost of purchasing AVs, for trip.
example. In poorer countries with lower wages, driver cost
is not as big a part of overall costs. The cost of infrastructure construction, when
amortized across all vehicles over time, is not a major
Public transport systems in this figure are shown with expense; certainly not in comparison with the cost of
drivers, though one can easily imagine the savings of drivers and, in the case of TNCs, the cost of operating
removing the driver. Even with drivers, public transport and managing the enterprise.
systems can provide the least expensive travel service of
any mode, though this is heavily dependent on average The cost per pkm for 2-wheel vehicles may appear
load factors. In the United States, with relatively low high especially for bicycles but this reflects the
average bus ridership, cost per pkm is higher than rail, relatively low average use of these vehicles in a given
while in India, with low driver wages and much higher year. A bike ridden 3 kilometers per day accounts
ridership levels, large buses clearly provide the lowest- for about 1,000 pkm per year, which results in higher
cost travel. overall costs per pkm than public transport modes.
The 20% overhead cost of operating and managing Costs change somewhat out to 2050, but the biggest
TNCs amounts to a significant expense; far higher impact on overall costs is how much these different
than the cost per pkm of operating public transport modes are used.
Figure 15. Cost per passenger kilometer by mode, 2030, U.S. and India examples
(Note: y axes are not aligned; Indian costs are far lower than U.S. costs)

Multiplying the per-pkm costs by total pkm in each scenario, the total costs of the scenario in a given year can be
obtained. These totals are shown in Figure 16, by scenario, worldwide to 2050. In 2030 the scenarios are remarkably
close together, reflecting in part that the revolutions are not that far along. It also reflects the fact that the significant
increase in shared trips in the 3R scenario occurs mainly with paid drivers, and does not provide a significant cost
advantage compared to private vehicle driving. However, by 2050, with widespread, lower cost automation and EVs,
and an ongoing increase in load factors across modes, the 3R scenario is far cheaper than the 2R scenario.
Figure 16. Total global cost of scenarios by year and mode
Again, these cost results do not include things like air pollution or traffic congestion impacts. They also ignore the value
of time, and the disutility of travel that may change considerably due to AVs or, for that matter, higher quality public
transport systems. But they do give a rough idea of the out-of-pocket costs to travelers and taxpayers associated with
building and operating, and using urban transportation systems. This also raises an issue that is beyond the scope
of this study: who pays? Some costs, such as the purchase and use of private vehicles or the fares paid for public
transport and on-demand mobility services, are borne by travelers. But other costs are paid by governments (and thus
taxpayers). These costs include highway infrastructure and often public transportation infrastructure. Thus savings in
each of these areas accrues to different types of agents; there would likely be winners and losers. All of these details
are beyond the scope of this analysis but would be areas for additional research.

5. Policy Narratives
The three scenarios we describe represent very different losses. In many non-OECD countries, low labor costs
futures. The course of events that cities and urban dampen demand for vehicle automation for many years.
transportation take will be highly dependent on the When demand increases, restrictions are enacted once
policies that city and national governments implement. AVs challenge labor markets.
Here we describe a set of policies that seems likely to BAU: Electrification

be needed in order to realize each scenario, and present
the scenarios roughly in order of the level of policy In the BAU scenario, electrification does not receive
intervention required. The BAU scenario, while perhaps additional government support to encourage widespread
increasingly unlikely due to its increases in travel, energy adoption, beyond what is already in place and planned. In
use and CO2 emissions, represents no significant change most OECD countries, this includes some modest support
in policy trajectory (and relatively limited success of the for EVs in the form of purchase incentives, new charging
new technologies in question); the 2R scenario would infrastructure, and some movement toward decarbonizing
require important policies to enable automation and electricity generation. These policies, however, have only
encourage electrification; and the 3R scenario would likely a limited impact and are not enough to help EVs break out
require the broadest array of policy interventions and of the 1-2% sales share position they are currently in. In
change as it affects fundamental mobility systems and most non-OECD countries little or no support is given to
patterns around the world. either the electrification of vehicles or the decarbonization
of electricity production.
BAU Scenario Policy Narrative
BAU: Sharing
The BAU scenario includes no major changes to the
course of current policies affecting urban transport New technologies permitting the sharing of trips continue
or land use. In the more advanced countries (United to receive some ongoing support in OECD countries.
States, Europe, China, Japan), this typically includes A few cities and countries continue to experiment with
continuing existing policies to support new technologies mobility-as-a-service platforms that better link public,
(electrification, sharing, and automation), urban private, and shared mobility services into a cohesive
development and transportation planning. In most network. However, these efforts are limited and do not
developing countries, the BAU scenario includes almost reach a large scale. Few governments actively encourage
no support for new technologies, limited capacity for the sharing of trips in any meaningful way beyond current
urban policy implementation, and generally continued policies supporting carpooling and public transport. In
policy support of (or failure to prevent) strong increases non-OECD countries, even less is done, and some policies
in private motorized travel, with transport infrastructure to actively restrict the sharing of trips via new technology
provision continuing to lag far behind urban growth. continue.
We describe the policy assumptions for BAU in some
detail here as these are likely to be just as important in BAU: Urban Planning
determining future directions as the new policy initiatives
we discuss for the other scenarios. Around the world, existing urban planning practices
continue into the future. In OECD countries in western
BAU: Automation Europe, and east Asia, existing policies that support
dense urban settlement tied to a strong network of
In OECD countries, most existing regulatory obstacles public transport, cycling, and walking continue. In more
to automation remain in place for many years. Even if car-centric OECD countries, such as those in North
automation technologies improve and mature, legal America and Australia, policies supporting low-density
liability, privacy, and ethics concerns are resolved only urban development tied primarily to automobile travel
slowly over the course of many years, and existing will continue, maintaining high mode shares of single-
restrictions on AVs are only removed after these issues occupant car travel over increasingly long distances.
are resolved. In some places, new restrictions on AVs In non-OECD countries, current urban planning trends
are enacted in response to growing concerns about job such as spending that is primarily focused on new roads,

flyovers, and limited access highways continues. Auto-centric and poorly-coordinated land use planning also continues
in many areas, leading to more low-density, vehicle-centric settlement patterns that follow large arterials as they
radiate out of cities. The speed of urban growth in non-OECD countries will continue to surpass transport infrastructure
provision. This, coupled with growing wealth and ability to purchase private vehicles, results in a large increase in
congestion and decrease in access. While a few countries will continue to invest in public transport and maintain
compact cities, most will not.
BAU Scenario: Assumed Policy Summary
Vehicle automation
* Maintain or slowly relax existing barriers to automation
* New restrictions added in some places to protect jobs
Vehicle electrification and decarbonization of electricity production

* Limited support for electrification and decarbonization in OECD countries
* Very low support for electrification and decarbonization in non-OECD countries
Shared mobility services

* Limited ongoing support for shared mobility, in select cities in OECD countries
* Low support for new vehicle sharing in non-OECD countries
* Continued support for compact cities, public transport, cycling, and walking in OECD countries where
this is already a focus
* Continued low-density car-centric development in the rest of the world
2R Scenario Policy Narrative Countries like Norway, with very high incentives to
purchase EVs, have shown that it is possible to achieve
The 2R scenario is focused on achieving rapid global high market shares (EV sales shares there reached 30%
adoption of EVs from 2020 onward, and a breakout and in 2016 (Lutsey, 2017). In major markets, ongoing strong
rapid growth trajectory for AVs beginning around 2025. purchase incentives must be accompanied by significant
This scenario is otherwise similar to the BAU scenario in use advantages (such as preferential parking) and an
the continuation of existing trends for vehicle sharing, ongoing increase in public and workplace charging.
public transport use, and urban planning. Many of the These types of incentives must continue until EVs become
measures described here would logically be undertaken market dominant and the costs of the incentive systems
at a national level; however some (such as restrictions on become unworkably high. Hopefully by that point the
ICE vehicles, coordination of EV charging infrastructure, tipping point toward a rapid increase in sales will be
etc.) would more logically be implemented by cities reached. Much of this cost of maintaining incentive
themselves. systems can be funded through differential taxation
systems, for example, feebates or French style bonus-
Electrification malus policies that tax the highest-emitting vehicles and
subsidize the lowest-emitting vehicles, which would
In both OECD and non-OECD countries, the short term include EVs (at least in countries with clean electricity).
push for electrification takes the form of policies that
reduce barriers to electrification and actively, often In the longer term, completely decarbonizing electricity
directly, support the decarbonization of electricity generation may require active closure of all carbon-
production. This includes government support and direct emitting power plants and their replacement with
provision of EV and clean power infrastructure. It also renewable power sources. Transitioning to a 100% EV
includes taxes and other market-driven schemes that fleet may require banning ICEs and government buyouts
restrict more polluting vehicles and power sources. of existing vehicle fleets.

Automation AVs become more cost competitive. When this happens,
these countries follow the lead of OECD countries in the
As mentioned earlier, the potential cost savings from development of policies that enable and support vehicle
vehicle automation makes it an attractive technology for automation.
businesses operating transportation services, especially
in high wage environments, such as in OECD countries. An important question is whether driverless cars are first
Thus the most crucial policies for automation to become and most significantly deployed in commercial fleets, taxi
widespread are those that remove restrictions to vehicle companies or TNCs, or in households. In this scenario,
automation. we assume that restrictions are removed for all potential
purchasers in an even manner and that, while there
The scenario envisions OECD countries leading the may be some early uptake commercially (given the high
creation of a policy environment that supports automation. commercial advantages of driverless cars), households
In addition to removing barriers to automation, are also quick to begin purchasing such vehicles. Thus
governments might also support research and testing by 2030 there is not a major difference in use patterns,
of automation technologies and set legal and safety except that intensively used commercial vehicles are
frameworks to allay public fears about personal data turned over much sooner, giving both more purchase
privacy and liability in the event of crashes. As labor prices opportunities to shift to AVs, and a faster penetration
rise and technology prices fall in non-OECD countries, through the stock.
2R Scenario: Assumed Policy Summary
Continuation of existing policies on sharing, public transport, urban planning (see BAU scenario)
No particular policy preference to increase EV or AV uptake by commercial operations over households
New policies supporting vehicle electrification and decarbonization of electricity production
* Carbon tax and heavy investment in very low carbon (e.g. renewable) electricity generation
* Subsidies to offset EV purchase costs; could be generated via cap-and-trade, feebates or other
market-driven schemes
* Elimination of subsidies for fossil fuels, making EVs more cost-competitive
* Require automated vehicles to be electric drive
* Support for public EV charging infrastructure
* Policies dedicating space for EV charging in cities
* Encourage smart charging of private and commercial EVs, at off peak times or otherwise in better
concert with electricity grid management systems
* Low emission zones and other policies to encourage EV use and/or restrict operation of ICE vehicles
in cities or their central business districts, or even broader bans on ICE vehicles
* Government research support for battery development and other new technologies such as
contactless charging to reduce charge time and increase driving range
* Government buyouts of polluting vehicles and subsidies for vehicle replacement
New policies supporting vehicle automation
* Remove major legal restrictions to AV use
* Develop comprehensive safety and liability regulations as consistently as possible across
jurisdictions
* Develop data policies to protect private travel data
* Support research into automation technology
* Set framework for AV testing and type approval
3R Scenario Policy Narrative

The 3R scenario builds on the 2R scenario, with policy support for both electrification and automation, but also
substantial policy support for shared-use mobility and urban planning that supports shorter trip lengths and high levels

of walking, cycling, and public transport use, even in a future where vehicular travel is significantly less expensive.
Without strong policy support for compact cities, even a scenario with fairly high levels of vehicle sharing in smaller
vehicles could result in significantly higher vehicle kilometers and lower levels of access.
TNCs must be encouraged to prioritize ride sharing over single-occupant rides, and to promote sharing as a preferred
option. Much of the 3R scenario outcomes, particularly in car-dominant countries, depend on the higher occupancy
rates of these cars. As cars become automated, and the per-trip costs drop, pricing incentives will likely become even
more important to encourage people to share trips. Zero-occupant trips by AVs (TNCs or private) must also be heavily
discouraged, probably through pricing, though the mechanisms for achieving this are not fully clear at this point.
Preference for AVs in public (shared mobility) rather than private hands is also likely to be important in this regard, to
avoid the higher driving levels of the 2R scenario.
Finally, policies will be needed to ensure that TNC vehicles work in concert with public transport and other highly
efficient modes, rather than compete with them. Policy options could involve possible restrictions on operations within
certain corridors, along with incentives to serve stations. Support for small bus and van programs that can provide on-
demand and at least near door-to-door service may also hold major potential for improving system efficiency.
3R Scenario: Assumed Policy Summary
Continuation of policies that support vehicle electrification and automation, as in the 2R scenario
New policies on EVs and AVs
* Discourage or restrict the operation of zero-occupant vehicles
* Discourage or restrict private ownership of AVs
Strong support for trip sharing, public and active transport
* Fees added for vehicular travel, or vehicle kilometers traveled, potentially variable to achieve the
desired level of movement, and with higher fees charged for vehicles with lower occupancies and
higher negative environmental and traffic impacts
* Conversely, vehicle kilometer subsidies could be applied to very high-occupancy vehicles (buses
and trains), particularly during high-congestion times on more congested routes
* Support and incentives for public transport operators to better match passenger demand with
vehicle size, through smaller automated electric vans and shared taxi-like vehicles
* Government support for driverless buses and rail, dramatically reducing the operating costs and
fares, while improving frequency and reliability for these shared modes
* As the nature of transit services changes, ensure mobility opportunities remain available and
affordable for traditional transit customers and for those with disabilities, older adults, and low-
income passengers
* Close attention to the labor and equity impacts of automation and shifts to shared mobility; ongoing
tracking and research into minimizing negative societal impacts of these revolutions
Policies on urban planning
* Mixed use, transit-oriented planning to encourage shorter, less car-dependent trips
* Better metropolitan area coordination of regional land-use and transportation decisions
* Increased, ongoing investments in walking, cycling, and public transport infrastructure and systems
* Improved safety as well as legal protection for walking, cycling, and public transit users
* Implementation of bike and e-bike sharing programs in urban areas with sufficient density
* Elimination of policies that increase motor vehicle use, such as minimum parking requirements, free
parking on public streets, and fuel subsidies
* Government coordination of mobility-as-a-service, linking many transportation options into a
seamless network of trip planning and payment via a single interface
* Increased use of local development impact fees; e.g. charges that account for car dependence and
other negative externalities, and these fees fund investment in sustainable transport
* Global institutions, such as development banks, change lending practices to shift investment from
urban roads toward public transport, walking, cycling, and other more sustainable modes

6. Conclusions, Uncertainties, Next Steps
The scenarios presented in this report are one way to would have large safety benefits, though these have not
construct alternative possible futures for urban passenger been quantified in this study.
transport. There are many other possible approaches.
This study attempts to identify how we can envision the To achieve the 3R scenario, pricing policies will likely
three revolutions in order to: need to play an important role. Such policies can help
avoid widespread use of zero-occupant vehicles and
Understand whether combining electrification with even single-occupant vehicles, and encourage enough
automation can deliver a low CO2 and otherwise on-demand mobility to get people around the world to
sustainable world cost-effectively shift away from private car ownership. The cost savings
associated with the 3R scenario are due mostly to
Understand whether further revolutionizing the manner factors that are not apparent and dont feature in much
in which we travel, rather than just the technologies decision-making today. For example, lower car ownership
on vehicles, can contribute significantly to achieving means people save enormous amounts on buying and
our environmental goals. maintaining vehicles that are unused at least 90% of the
time.
Ultimately the findings shown here are not that surprising.
While automation may produce some efficiencies and Similarly, savings from fewer roads and parking lots may
road congestion benefits, it may also trigger increases only become apparent as savings opportunities once the
in travel, and by itself does not seem likely to result in a world is strongly on a 3R path.
low-energy future. But when paired with electrification,
the combination does seem capable of providing deep Overall, while our pathway to a revolutionary future
energy and CO2 cuts (and thus nearly all of these cuts, on appears to have low direct costs and likely high societal
balance, can be attributed to electrification). Even then, benefits, it may require aggressive, visionary policymaking
the net effects of electrification depend on decarbonizing to achieve policymaking to achieve.
electric power.
We also find that there are fairly enormous efficiencies

associated with shared mobility, including ride sharing
in LDVs along with the use of small and large buses and
urban rail systems. Shared mobility in ride-hailing vehicles
must achieve relatively high average occupancy (load
factor) levels in order to really contribute to reductions
in traffic and energy use; for their part, public transport
modes must provide efficient, comfortable service for
this 3R scenario to succeed. There is evidence that such
high quality public transport is achievable, but there are
no guarantees and considerable policy pressure and
investments will be needed.
If successfully achieved, a 3R scenario with its ecosystem

of shared vehicle trips, public transport, and active travel
use can provide high quality and sufficient urban mobility
(indicated in this study as pkm) with far fewer vehicles on
the road, and even fewer vehicles parked, compared to
our BAU or 2R scenarios. Energy and CO2 emissions in
the 3R scenario are about half those of the 2R scenario,
and costs perhaps surprisingly are far lower as well.
The 3R scenario also seems very likely to provide the
biggest benefits in cutting traffic congestion and likely

Bibliography
Alonso-mora, J., Samaranayake, S., Wallar, A., Frazzoli, E., & Rus, D. (2017). Trip-Vehicle Assignment.
https://doi.org/10.1073/pnas.1611675114/-/DCSupplemental.www.pnas.org/cgi/doi/10.1073/pnas.1611675114
Anair, D. (2017). Capturing the Climate Benefits of Autonomous Vehicles.

Retrieved from https://3rev.ucdavis.edu/wp-content/uploads/2017/03/3R.Climate.Indesign.Final_.pdf
Automated Driving: Levels Of Driving Automation Are Defined In New SAE International Standard J3016. (2016). Global
Ground Vehicle Standards.
Retrieved from https://www.sae.org/misc/pdfs/automated_driving.pdf
Beiker, S., & Meyer, G. (2014). Disruptive Innovation on the Path to Sustainable Mobility. Road Vehicle Automation.
https://doi.org/10.1007/978-3-319-05990-7
Brown, A., Gonder, J., & Repac, B. (2014). An Analysis of Possible Energy Impacts of Automated Vehicle. Road Vehicle
Automation, 6170. https://doi.org/10.1007/978-3-319-05990-7
Campbell, A. A., Cherry, C. R., Ryerson, M. S., & Yang, X. (2016). Factors influencing the choice of shared bicycles and
shared electric bikes in Beijing. Transportation Research Part C: Emerging Technologies, 67, 399414.
https://doi.org/10.1016/j.trc.2016.03.004
Cohen, S., & Shirazi, S. (2017). Can We Advance Social Equity with Shared, Autonomous and Electric Vehicles?
Retrieved March 20, 2017, from https://3rev.ucdavis.edu/wp-content/uploads/2017/03/3R.Equity.Indesign.Final_.pdf
Davies, A. (2015). Turns out the hardware in self-driving cars is pretty cheap. Retrieved from https://www.wired.com/2015/04/
cost-of-sensors-autonomous-cars/
Elgowainy, A., Han, J., Ward, J., Joseck, F., Gohike, D., Lindauer, A., Wallington, T. J. (2016). Cradle-to-Grave Lifecycle
Analysis of U.S. Light-Duty Vehicle-Fuel Pathways: A Greenhouse Gas Emissions and Economic Assessment of
Current (2015) and Future (2025-2030) Technologies.
Fagnant, D. J., & Kockelman, K. M. (2013). Preparing a nation for autonomous vehicles. Transportation Research Part A
Journal, 77(October), 132. https://doi.org/10.1016/j.tra.2015.04.003
Fulton, L. M., Seleem, A., Boshell, F., Salgado, A., & Saygin, D. (2017). Electric Vehicles: Technology Brief.
Garrick, D. (2016). Car2Go switching electric cars to gas. Retrieved March 20, 2017, from https://www.iea.org/topics/
transport/subtopics/mobilitymodelpartnership/
Googles Autonomous Vehicle. (2012). Retrieved March 23, 2017, from http://googlesautonomousvehicle.weebly.com/
technology-and-costs.html
Handy, S. (2017). Active Transportation in an Era of Sharing, Electrification and Automation. Retrieved from https://3rev.
ucdavis.edu/wp-content/uploads/2017/03/3R.Active.InDesign.Final_.pdf
IEA. (2016). Energy Technology Perspectives 2016 Towards sustainable urban energy systems. Executive Summary.
Retrieved March 20, 2017, from https://www.iea.org/etp/etp2016/
IHS Markit. (2014). Self-Driving Cars Moving into the Industry's Driver's Seat. Retrieved March 23, 2017, from http://news.
ihsmarkit.com/press-release/automotive/self-driving-cars-moving-industrys-drivers-seat

Johnson, C., & Walker, J. (2016). Peak car ownership. Rocky Mountain Institute.
Lutsey, N. (2017). The rise of electric vehicles: The second million.

Retrieved March 1, 2017, from http://search.proquest.com/docview/1266038343?accountid=14549%5Cnhttp:hl5yy
6xn2p.search.serialssolutions.com/?genre=article&sid=ProQ:&atitle=From+the+blogs&title=Financial+Management&i
ssn=14719185&date=2012-11-01&volume=&issue=&spage=16&author=Barman,+T
Mason, J., Fulton, L., & McDonald, Z. (2015). A Global High Shift Cycling Scenario: The Potential for Dramatically
Increasing Bicycle and E-bike Use in Cities Around the World, with Estimated Energy, CO2, and Cost Impacts.
Institute for Transportation & Development Policy.
McFarland, M. (2015). The $75,000 problem for self-driving cars is going away.
Retrieved March 1, 2017, from https://www.washingtonpost.com/news/innovations/wp/2015/12/04/the-75000-
problem-for-self-driving-cars-is-going-away/?utm_term=.1d1814825145
McKerracher, C., Itamar, O., Wilshire, M., Tryggestad, C., Mohr, D., Hannon, E., Moeller, T. (2016). An integrated perspective
on the future of mobility. Retrieved from https://www.bbhub.io/bnef/sites/4/2016/10/BNEF_McKinsey_The-Future-of-
Mobility_11-10-16.pdf
OECD - International Transport Forum. (2015). Urban Mobility System Upgrade: How shared self-driving cars could change
city traffic. Corporate Partnership Board Report, 136. https://doi.org/10.1007/s10273-016-2048-3
Ory, D. (2017). Governance: Whos in Charge Here? Retrieved from https://3rev.ucdavis.edu/wp-content/uploads/2017/03/3R.

Governance.Indesign.Final_.pdf
Polzin, S. E. (2017). Three Transportation Revolutions: Synergies with Transit. Retrieved March 20, 2017, from https://3rev.
ucdavis.edu/wp-content/uploads/2017/03/3R.Transit.Indesign.Final_.pdf
Replogle, M. A., & Fulton, L. M. (2014). A Global High Shift Scenario: Impacts and Potential for More Public Transport,
Walking, And Cycling With Lower Car Use. Institute for Transportation & Development Policy.
Schaller, B. (2017). Unsustainable? The Growth of App-Based Ride Services and Traffic, Travel and the Future of New York
City. Retrieved March 20, 2017, from https://www.iea.org/topics/transport/subtopics/mobilitymodelpartnership/
Shaheen, S., Chan, N., Bansal, A., & Cohen, A. (2015). Definitions, Industry Developments, and Early Understanding.
Retrieved from http://innovativemobility.org/wp-content/uploads/2015/11/SharedMobility_WhitePaper_FINAL.pdf
Wadud, Z., MacKenzie, D., & Leiby, P. (2016). Help or hindrance? The travel, energy and carbon impacts of highly automated
vehicles. Transportation Research Part A: Policy and Practice, 86, 118. https://doi.org/10.1016/j.tra.2015.12.001

Itdp 3r Report Final

Uploaded by

Copyright:

Available Formats

Itdp 3r Report Final

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Itdp 3r Report Final

Uploaded by

Copyright:

Available Formats

Three Revolutions in Urban

Lew Fulton, UC Davis

Research supported by:

SUSTAINABLE TRANSPORTATION ENERGY PATHWAYS

i Three Revolutions in Global Transportation

ii Three Revolutions in Global Transportation

Each revolution addresses different societal needs, but

1 Three Revolutions in Global Transportation

2 Three Revolutions in Global Transportation

3 Three Revolutions in Global Transportation

The 3R scenario will require strong additional

4 Three Revolutions in Global Transportation

Shared vehicle trips While public transport

These three revolutions are highly uncertain in many

5 Three Revolutions in Global Transportation

Synergies Achieved by Combining Revolutions

Electrification can assist in the power and electronic demands of AVs.

7 Three Revolutions in Global Transportation

8 Three Revolutions in Global Transportation

9 Three Revolutions in Global Transportation

Table 1. Examples of AVs in development

Company Vehicle Brand Model Powertrain Production Notes

GM Chevrolet Bolt Electric Testing 40 cars in

FCA Chrysler Pacifica Hybrid Plug-in Hybrid Testing 100 vehicles

Ford Ford Fusion Hybrid 2021

Volvo Volvo XC90 Hybrid

Uber Ford Fusion Energi Plug-in Hybrid

Uber Volvo XC90 Hybrid

Daimler Mercedes- F 015 Luxury in Hydrogen Fuel Research Vehicle

Hyundai Hyundai Tucson Hydrogen Fuel Testing 2 vehicles in

10 Three Revolutions in Global Transportation

Figure 2. Potential changes in energy consumption due to vehicle automation

11 Three Revolutions in Global Transportation

12 Three Revolutions in Global Transportation

13 Three Revolutions in Global Transportation

Increased use of shared mobility systems, public transport

14 Three Revolutions in Global Transportation

Table 2. Key characteristics of the 3 Revolutions scenarios

Use of Use of Use of Shared Urban Aligned with

BAU, limited Low Low Low Low No

2R with high High High Low Low Maybe

3R with High High High High Yes

The BAU Scenario

15 Three Revolutions in Global Transportation

16 Three Revolutions in Global Transportation

Assumptions for Automated EVs

17 Three Revolutions in Global Transportation

18 Three Revolutions in Global Transportation

19 Three Revolutions in Global Transportation

20 Three Revolutions in Global Transportation

Results: Passenger Travel Projections Across the Scenarios

21 Three Revolutions in Global Transportation

22 Three Revolutions in Global Transportation

Vehicle Sales and Stocks

23 Three Revolutions in Global Transportation

Figure 10a-d. Evolution of vehicle and pkm of travel, U.S. example

24 Three Revolutions in Global Transportation

Scenario Impacts on Energy and CO2 Emissions