work as a ‘shock
times when the
demand for travel
is very high.”
DITCHING THE DRIVER
The third emerging technology is self-driving cars. When self-driving cars take
hold, Ostrovsky says, they will magnify
the benefits of the other two technologies.
For one thing, a carpooling system based
on robo-taxis would be more dependable
and reliable than one with human carpool
drivers. For instance, if a commuter
carpools to work but then the system has
trouble finding a convenient carpool match
for her return trip home, the system can
fall back to having the commuter ride solo,
for a part or for all of the trip. And this
“fallback” can be done seamlessly for the
commuter, saving her the stress and hassle
of arranging alternative transportation.
Self-driving cars can also greatly reduce
various frictions involved in forming
carpool matches and increase flexibility in
the types of carpool matches that can be
made. With a self-driving car, for example,
the first person to be picked up does not
have to also be the last one to be dropped
off. The car can go anywhere, mapping
routes based on its mix of customers. It also
becomes easier to make carpool matches in
real time, adjusting the route on the fly.
More broadly, says Ostrovsky, a system
based on self-driving cars, carpooling,
and tolls will blur the line bet ween
public transportation and solo driving,
giving commuters access to convenient,
reasonably priced door-to-door transit with
a small number of stops in the middle.
There is another reason why the arrival
of self-driving cars may make tolls and
carpooling necessary. Self-driving cars
will make commuting easier and more
comfortable — for instance, a rider will
be able to sleep in the car. This will make
longer commutes tolerable. Thus, the
arrival of self-driving cars will increase the
number of cars on the road.
On the flip side, it is also expected that
self-driving technology will lead to an
increase in road throughput. The overall
impact on traffic congestion is ambiguous,
but there is a serious danger that traffic
will get much worse. In that case, tolls will
be necessary to discourage commuters
from long solo drives during rush hour.
And when tolls are introduced, carpooling
becomes more attractive, because it allows
riders to share the costs.
Ostrovsky and Schwarz analyzed these
new dynamics to determine best
principles for future policy and developed
a computationally tractable methodology
that could be used to set road prices. Their
analysis also produced several general
First, you do not need to impose a charge
for every road at every moment of the day.
You only need to impose a charge when
demand is above the road’s capacity. If the
traffic is flowing smoothly even without
a charge, there is no benefit to charging
a price that reduces the traffic even more.
In fact, charging a price for an uncongested
road would actually reduce incentives for
people to drive at times when traffic is lower.
Second, you do not need to
micromanage every detail. If you set
appropriate road tolls and have a well-
functioning market for carpooling, there is
no need for regulators to force commuters
into arrangements they don’t want. The
carpooling decisions that people make in
their own self-interest will end up being
good for the city as a whole.
However, it is not a good idea to leave
everything in the hands of market forces.
If a road’s prices are set by a private
company, for example, the owners would
want to collect tolls all day and every
day, including the times when roads are
underutilized. While this would be great
for the private company’s bottom line,
such tolls would be detrimental to the
commuters and reduce overall efficiency.
To be of maximum benefit, the
researchers warn, road pricing should be set
by a public authority with a city’s broadest
public interest in mind.
“Traffic congestion is a major issue
worldwide, imposing a serious burden on
individuals and families and causing a host
of other problems, such as wasted fuel
and environmental damage,” Ostrovsky
says. “We hope that the combination
of intelligent road pricing, convenient
carpooling, and eventually autonomous
transportation will help solve this problem,
and we view our proposed framework as
a step toward the solution.” Δ
But carpooling is on the brink of its own
revolution, Ostrovsky says. Just as Uber and
Lyft made it easy to match for-hire drivers
with passengers, new services like Waze
Carpool and Scoop are starting to make
it easy for commuters to match with each
other and organize carpools. The new
platforms can match passengers heading
to similar places, calculate the fastest
or cheapest routes, and handle the billing.
In time, carpooling could be almost
as cheap as riding a bus and almost as
convenient as driving solo.
A crucial point, Ostrovsky says, is that
flexible carpooling will also make
congestion pricing much easier to tolerate.
It allows people to share the cost, thus
reducing the financial burden of tolls.
Commuters could also choose to drive solo,
but would then pay the entire cost of tolls.
Those opportunities reinforce the
benefits of congestion pricing, because
people can choose the way they travel based
on their individual preferences about cost
and convenience. If congestion pricing
creates a more efficient market for traffic,
carpooling can increase the efficiency even
more. Carpooling gives commuters an
opportunity to travel during times of high
demand without being forced to pay high
prices. As Ostrovsky puts it, “Intelligent
tolls and convenient carpooling, deployed
together, will work as a ‘shock absorber’ for
times when the demand for travel is very
high, leading to a higher average number of
people per car instead of congested roads.”