Free, open-access streets and highways in dense areas tend to get overrun by cars and other vehicles, causing frequent, severe traffic delays and reducing throughput. Carpooling allows more people to use roads, but the incentives to carpool are often too weak; despite saving travelers money, carpooling does not save them time. Dynamic tolling, where tolls adjust according to demand in order to keep traffic flowing quickly, dramatically reduces traffic delays and boosts highway capacity. It also incentivizes carpooling, because carpoolers can split the cost of tolls. This little-discussed benefit allows for more people to make journeys more quickly on existing road infrastructure. Autonomous vehicles (AVs) will increase travelers’ comfort and lower the costs of operating a car, discouraging carpooling and increasing people’s willingness to endure traffic delays. In very busy areas, traffic delays will likely increase. Dynamic tolling is a key tool in promoting smooth traffic flow and incentivizing efficient carpools, and as AVs become more common, its importance will grow.
Why Dynamic Tolls Promote Carpooling
Carpoolers can split the cost of gas. They can also roughly split the per-ride costs of insurance and wear and tear. With app-hailed ride-shares, such as Uberpool, Lyft Line, and Via, passengers can also share the cost of a private driver. However, sharing a ride does not affect what on many journeys is the largest cost: traffic delays. People travelling together will endure the same time costs as everyone else: on a congested road, they move just as slowly as people who are driving alone.
Dynamic tolls are key to fighting congestion. Dynamic tolls discourage drivers from travelling on overwhelmed streets and highways, reducing congestion and speeding up journeys. The more people want to use a road (or a special tolled lane), the higher the tolls go, discouraging additional drivers from taking that route. This means that vehicles, including buses, can nearly always travel quickly. Congestion pricing not only speeds up journeys, but also increases the volume of cars that can use an otherwise congested highway. More cars can use a road if they are going faster; when traffic overloads a highway’s maximum capacity (generally about 2,000 cars per lane per hour at 55mph/89kph), vehicles slow down, allowing fewer people to get through. Congestion pricing keeps traffic moving by ensuring that this maximum capacity is never exceeded.
Dynamic tolls also introduce new incentives for carpooling. When the “price” of using a busy highway is an extra 45 minutes sitting in traffic, carpooling does nothing to reduce it. The 45-minute delay merely becomes a 45-minute delay with company. When the price to avoid that delay is a $10 toll, two people carpooling can each chip in $5, half of what they would when driving alone. Financial costs can be split; time costs cannot. The more people there are in the vehicle, the lower the per-person cost; in a 10-passenger commuter van, each rider would only pay $1 extra to use the dynamic toll lanes. Plus, the toll lanes would save those commuters money as well as time, as they would have to pay for 45 fewer minutes of a hired driver’s time (e.g. $15/hr*.75hrs=$11.25).
Surface streets and high-traffic areas of cities can also benefit from dynamic tolling. Reducing the number of cars on overwhelmed streets would allow traffic to move quicker, and vehicle-sharers would save money while travelling much faster than they do now. Though traffic on local streets is harder to target than on highways, dynamic tolls/congestion fees (and flat-rate charges less so) can help substantially improve flow. One option is to establish dynamic toll lanes or a tolling system for all non-local traffic on specified arterial roads. Another option is to set a cordon around an area or areas (generally the central business district) and toll drivers by how much time they spend in that area. This system is often applied with a flat-rate fee, as opposed to a dynamic one. While flat-rate fees do discourage driving in a tolled area, they are often too low to control rush-hour traffic flows and might be unnecessarily high for mid-day travel. A dynamic toll would adjust based on demand and thus better fight traffic. A GPS-based city-wide tolling system would raise serious privacy and practicality concerns, but would offer more fine-grained targeting of congestion.
The Role of Ride-Sharing Platforms
Dynamic tolling goes a long way to encourage carpooling, but its effects will be much stronger if the coordination costs of carpooling are reduced. Virginia’s I-66 high-occupancy toll (HOT) lanes have over 50% carpool share (2+ people) during rush hours, but these lanes let carpools travel for free. Carpools of 3+ people in HOT lanes have proven rare in cities such as Atlanta, even when three is the minimum occupancy for free travel. Carpools have traditionally been very difficult to coordinate, requiring two or more people know and trust each other, consistently coordinate their schedules, show up on time, and rarely cancel. For many people, this isn’t worth the hassle, even when there are strong financial incentives.
Technology can help lower carpooling coordination costs dramatically. Over the past five years or so, platforms have emerged that allow people to coordinate carpools that do not rely on a specific driver but instead match strangers who want rides with drivers who will take them. BlaBlaCar matches people for long-distance, city-to-city trips, and its newer BlaBlaLines service matches them on select routes for short trips. Lyft Line, UberPool, and Via have also made it easier for people to share rides. These platforms as well as new entrants stand to innovate and improve the carpooling process further, and still new innovation. I have a post that dives a lot further into potential innovations in coordinating carpools.
Dynamic tolling increases technology companies’ incentives to lower the coordination costs of carpooling by making their sharing platforms more efficient. It creates new financial incentives for people to carpool, and modern platforms/apps can help people create carpools. Factors such as passengers’ average wait time, the size of vehicles, and the number of stops would be influenced by prices; the most sensible kind of service will emerge in any given situation. For instance, a road with very high tolls (due to high demand) would encourage a greater number of large carpools (e.g. 4+ people) than a road with lower tolls.
Dynamic tolls will impose a fair cost on all private vehicles using busy roads. Ride-sharing is often blamed for clogging up city streets and worsening urban environments. Cities such as Chicago and New York have set quotas for ride-share vehicles in an effort to improve the traffic environment. There is now evidence that these services have indeed increased traffic in busy areas. However, ride-share cars comprise just a fraction of the vehicles on the road; the real traffic problem comes back to the underpricing of streets and highways. Many cities used blunt instruments such as parking restrictions and taxi licensing to temper traffic, but even before the arrival of Uber and company, these measures have not been nearly enough to relieve traffic in dense areas. If road space for cars were instead allocated using prices, cities could finely control traffic flows and prevent gridlock. All vehicles would have to pay for the road-space they use, and the result would be a reduction in the number of ride-share vehicles, but also in the number of personal automobiles. This solution is fair to all parties, and unlike imposing quotas on ride-shares, it encourages shared rides of all kinds. Every private automobile should pay its fair share to use busy streets, and if this happens, carpooling will follow.
Why a Future with Autonomous Vehicles Needs Dynamic Tolling
Autonomous vehicles stand poised to revolutionize transportation, but road pricing will be key to ensuring that they make cities more useful and livable. AV hailing platforms (apps) will attract many more customers than their human-driven versions. They’ll offer cheaper rides, more demand-responsive service, and better algorithms for shared rides. AVs will make more efficient use of roadspace, but this will simultaneously encourage people to take more journeys by car (a phenomenon known as induced demand), and many busy areas will remain overrun with vehicles. The lack of driver costs and the many other efficiencies offered by AVs will mean that private rides will be cheaper, and sharing will save people less money. Say AVs cut a given ride’s cost (on un-tolled roads) from $35 to $20; splitting that cost with another person then means saving $10 instead of saving $17.50. Moreover, as car travel gets cheaper and more comfortable, people will be willing to spend more time sitting in traffic. They won’t have to spend as much on gas, and they’ll be able to watch movies instead of the road. AVs will also be able to cheaply deadhead (go without passengers) toward destinations where they could get more fares, which will mean even more vehicles will be on the road at any given time. However, with road pricing, the number of AVs prowling busy streets and highways wouldn’t surpass the designed maximum, encouraging people to share when it makes sense and fostering a stable equilibrium of smoothly-flowing traffic.
Large fleets of autonomous vehicles, frequent use of ride-sharing platforms, and smart demand-estimation algorithms can together make carpooling easier than ever, and road pricing would ensure that this happens. Average vehicle occupancy, along other aspects of the AV experience, would emerge naturally as a response to consumer demand and road prices. The average carpool on the highest-demand journeys might include 8 people but only require one or two stops to drop all of them off. The average occupancy for lower-demand trips might be just one or two people. In both cases the system is working; in the latter, most people don’t think carpooling is worth it. Carpooling is a means to the goal of better mobility; it should happen whenever it makes sense to people. However, without tolls, the average vehicle occupancy on all roads, even those with heavy traffic, would likely be lower than it is today, and congestion would remain an issue in busy areas.
Dynamic tolling is the most elegant solution to traffic and the best way to encourage sensible carpooling. Whether humans or computers are driving, tolls can be shared, but time costs cannot. Dynamic tolls can stop congestion and promote smart, shared mobility; they help cities and regions move forward.
Postscript: Express Toll Lanes vs High-Occupancy Toll Lanes
Some highways with dynamic toll lanes offer discounts for carpools with 2+ or 3+ passengers (high-occupancy toll or ‘HOT’ lanes), but in express toll lanes (ETL), i.e. those without discounts to carpools, everyone pays, and sharers split the costs on their own. In ETLs, passengers cannot lie about how many people they have in the car. The toll is charged to the vehicle itself; someone driving alone will bear the full cost, and carpoolers will split the cost amongst themselves. The incentive to carpool is not as high, but the system is simpler, and the tolls that are charged will be lower, all else equal. Highway operators do not have to monitor whether the vehicles claiming discounts are actually carpools, as is required with HOT lanes. In some cases, people have to register their carpool to qualify for a free or discounted ride, many people end up paying the full toll because they haven’t registered. HOT lanes also fail to incentivize carpools larger than the minimum; using the lanes is free whether a vehicle has three people or ten. With simple express toll lanes, there is more incentive for people to choose exactly how and when they want to share, as opposed to just choosing whether or not to share.
One Last Thing: The author hereby calls dibs on the name “AV Club” for an app for hailing autonomous vehicles.
Rapid Transient 
5 Comments