Transportation Network
Companies

Transportation Network Companies (TNCs) like Lyft, Uber and other similar services are increasingly likely to fundamentally alter travel behavior at the local and regional levels. TNCs allow a person to easily obtain point-to-point rides through smart phone interfaces with integrated payment systems.

 

Transportation Network Companies (TNCs) like Lyft, Uber and other similar services are increasingly likely to fundamentally alter travel behavior at the local and regional levels. TNCs allow a person to easily obtain point-to-point rides through smart phone interfaces with integrated payment systems.

For a variety of reasons, TNCs have become controversial, with local, regional and state agencies responding to their proliferation in a variety of ways. In some locations TNCs have been enthusiastically welcomed, while other communities have pursued legal action to limit their operations. One oft-cited potential benefit of TNCs is their ability to reduce vehicular travel, providing associated air quality, fuel use, and greenhouse gas benefits. Other potential benefits of TNCs include:

  • Reducing the rate of vehicle ownership, similar to existing car sharing services
  • Providing convenient “first/last mile” access to transit, like Centennial, Colorado is doing
  • Allowing more efficient use of existing vehicles
  • Allowing employers, agencies, and organizations to outsource their transportation programs

While many commentators expect TNCs will reduce vehicular miles traveled (VMT), it is also possible that, due to their convenience and relatively low cost, TNCs could actually induce additional travel or shift trips away from low-impact transit, bicycling or walking modes. Researchers, public agencies and TNC firms themselves are beginning to explore these issues using a variety of methods. To date, there have been several studies released on this topic. However; much of this research is limited either in terms of time period, geography, types of service, data availability and other factors. Additionally, there may be long-term behavior effects requiring more extensive study to fully understand.

While it is difficult for transportation professionals and stakeholders to draw definitive conclusions at this time using the available data, we have outlined three potential scenarios which address possible outcomes:

  1. TNCs decrease vehicular travel
  2. TNCs actually increase vehicular travel
  3. We don’t yet know definitively, but here is an interim strategy

Though TNCs are moving quickly to enter many urban markets, it may take years to discover their full effect on travel behavior.  In the meantime, we are in active discussions with local governments, the TNCs themselves and related businesses to better understand the early evidence on how they are performing, the conditions that are being imposed on new services, and the underlying elements of traveler choice that are leading to their use and resulting effects on other modes and vehicle miles traveled.

Will these ultimately decrease or increase vehicular travel? We’re exploring three possible scenarios based on active discussions with TNCs and local governments:

TNCs will decrease vehicular travel...

Under this argument, there are several characteristics of TNCs that could result in decreased vehicular travel. The factors through which TNCs could result in reduced VMT include:

    • Reduction in Trip Lengths – Part 1: While there may not be consensus on the actual proportion, many TNC trips replace trips formerly taken by taxi. Taxi drivers often circulate along heavily traveled routes looking for fares. TNCs make ride scheduling more efficient, where drivers can position themselves to optimize ride requests. Thus, the total miles driven without a passenger (or “deadheading”) is potentially reduced compared to taxis.
    • Reduction in Trip Lengths – Part 2: In dense urban areas where the service reaches a critical mass of users such that between-ride TNC cruising and deadheading is minimized and, in areas with scarce parking, TNCs eliminate the amount of circulating self-drivers need to perform in search of a parking space. This means trip lengths are more direct, resulting in less driving.
    • More Efficient First/Last Mile Connections: To provide convenient access to transit stations where limited parking deters transit use or where travel to/from the station are balanced at peak times of day, TNCs can carry travelers both to and from the station. More efficient first/last mile connections could also encourage some mode shift from driving alone to transit, as commuters may be more likely to leave their car at home if they can take transit combined with a seamless last mile TNC connection instead.
    • More Ridesharing: When shared-ride demand for services such as UberPool and Lyft Line reach a “tipping point” such that single trips can accommodate multiple groups’ origins and destinations with minimal out-of-direction travel, total VMT may decrease. Uber recently published a blog post which claims that UberPool in San Francisco reduced VMT by 674,000 miles across a one-month period, relative to individual Uber rides (they did not compare to individual drivers or trips by transit).
    • More Efficient On-Demand Transit Service: In sparse areas with life-line transit service, where low-ridership bus routes can be replaced with on-demand TNC services, only trips specifically requested will occur. Other unneeded trips do not occur, reducing total vehicle travel.
    • More Efficient Employer Carpools: When business-focused TNC services such as Lyft for Work can replace less effective employer-operated ride-share programs, travel may decrease. Under this scenario, appealing on-demand ride service is provided exactly at the right level to match high usage demand generated at concentrated employment locations.
    • Synergies with Other Delivery Services: In the future, at a point when package delivery services could be integrated with traveler services in a way that preserves passenger convenience while making productive use of miles that would otherwise be generated cruising or deadheading between rides, total system VMT could decrease.

Considering for the factors listed above, it remains unclear if these factors – when taken together – will add up to less vehicle travel than would have otherwise occurred if TNCs were not in existence.  TNCs provide a more convenient alternative to some who would otherwise walk, cycle or use transit, so any shifts in mode choice are likely to move those willing to pay for a TNC service into modes that generate a higher number of vehicle miles per person-mile of travel.  Those unwilling to pay are likely to remain in their current modes.

Conversely, those who shift from driving a car to using a TNC are not likely to reduce VMT.  Self-drivers move directly from origin to destination and then park, while TNCs – to varying degrees – generate unproductive VMT as they cruise or deadhead to serve their next ride.  This argument is described under Option 2 below.

TNCs will increase vehicular travel...

Even if TNCs will reduce the number of cars on the road, the convenience and availability of them will make them a highly attractive travel option. This means that any shift from non-auto modes will result in greater VMT.  And, while there is a reduction in VMT for shared-ride TNC trips, there is still added VMT to divert off-route to pick up passengers, or repositioning of vehicles (possibly “deadheading”) to pick-up locations.

Going further, some primary reasons why TNCs might increase VMT are:

    • Convenience: TNCs provide a more convenient alternative to some who would otherwise walk, cycle or use transit, so any shift in mode choice is likely to move those willing to pay for the service into modes that generate a higher number of vehicle miles per person-mile of travel. This is especially true in cities like San Francisco, where low-cost shared ride services like UberPool and Lyft Line are price-competitive with transit and have already caused some mode shift away from transit.
    • Trip Directness: This is the counter argument to the “reduction in trip lengths” points described above. Even those who shift from driving a car to using a TNC may not generate a reduction in VMT when considering the entire trip. Self-drivers move directly from origin to destination and then park, while TNCs to varying degrees, generate unproductive VMT as they cruise or deadhead to serve their next ride.  We see this as an important area of research to better understand the extent to which unproductive VMT generated by TNCs offsets their shorter trip lengths.
    • Induced Travel: There may be instances in which a person makes a trip that they otherwise might not make, particularly in dense urban areas where TNCs are readily accessible with limited delay.  History has shown that reductions in the cost of travel (gas prices, parking prices) or increases in the convenience of travel (reduction in congestion, capacity increases) tend to increase the amount of travel.  By allowing a person to obtain a ride through a simple swipe on their Smartphone, there is the potential that travelers may make additional trips because traditional barriers such as congestion and parking are reduced or even eliminated through the use of a TNC.

There has been a small sample of quantitative studies that evaluated the TNC’s effect on travel behavior including:

    • Research at the University of Texas by Fagnant and Kockelman modeled shared-ride vehicles and found that VMT increased by 1.5 to 4.5% for dynamic ridesharing, depending on the extent to which dynamic ridesharing replaced single auto trips. This study applied a simulation process to evaluate how travel behavior might change under the widespread deployment of shared use vehicles.  In this particular instance, the shared use vehicles were assumed to be autonomous.

A recent study by the International Transport Forum found that while the number of vehicles decreased, travel itself increased, showing a 6-89% VMT increase, depending on the extent to which TNC’s replace high-density transit service.

We don't know yet definitively, but here is an interim strategy...

Given the uncertainty about the short-term and long-term effects of TNCs, one option would be to use various analytical tools to systematically test potential outcomes by varying input data.   This approach would then identify a range of potential end states, which could then inform stakeholders, transportation planners, and other interested parties.

One readily available analytical tool is a Citywide or Regional travel demand model.  These models are typically used to evaluate the effects of land use and transportation network changes within a jurisdiction.  In this instance, various inputs would be modified to replicate behavioral changes that may occur with the widespread use of TNCs.  Variations in the travel model outputs, such as vehicle miles traveled (VMT), could then be observed.

While many variables could be considered, several key input variables that could be adjusted include:

    • Auto Ownership: Reducing auto ownership levels, either within a defined area (such as an urbanized downtown) or uniformly throughout a region. This change simulates what might happen with the widespread use of TNCs resulting in a reduction in automobile ownership.    This scenario is likely to reduce vehicular travel, depending on the exact assumptions used.
    • Travel Costs: The cost of travel could be decreased, which would reflect ready access to TNCs. This approach simulates a condition in which TNCs are meeting demands for unmet travel because of the greater convenience offered by their service.  We would expect VMT and other travel metrics to increase under this scenario.
    • Transit Ridership: Transit ridership could increase, reflecting greater usage of TNCs to access transit and also to address first-mile/last-mile issues. This change would likely reduce vehicular travel.  Alternatively, transit ridership could decrease and VMT could increase, as TNCs provide higher quality of door-to-door service for those willing to pay more.
    • Parking Costs: The cost of parking could be reduced, which would reflect a condition in which a traveler could use a TNC to reduce the need to find parking within selected areas. This approach would likely increase vehicle miles traveled.
    • Ridesharing: To the extent that ridematching services offered by TNCs catch on, more carpool trips could occur, leading to higher average vehicle occupancy levels and reducing regional VMT.
    • Land Use Relationships: Similar to past research efforts by Fehr & Peers (and others) demonstrating how built environment factors influence travel behavior, we expect that TNC demand and travel patterns also depend on “Ds” factors such as density, diversity, destination accessibility and distance to transit.
    • New Mode Choices: At a more technically advanced level, models could be updated to include a specific TNC and/or microtransit mode, with specifically defined user characteristics that could be updated and refined as better survey data on current travel mode choices becomes available.

Under this approach, a nested mode choice model would need to be used to test whole-trip trade-offs in terms of time, cost, reliability, comfort and convenience of TNC service compared with all other modes.  Traffic routing and assignment models should be used to account for the unproductive VMT generated by TNCs that are circulating in wait of their next ride or deadheading back to high ride activity areas after a drop off in a less intensive area.  Finally, the process should also account for the ability for TNC ride-sharing services such as UberPool and Lyft Line to minimize these inefficiencies.