How Public Transit And Roads Are Different

We must think about the capacity and functionality of each.

Lines on a map are not capacity.

Public transit and public roads are very, very different.  A common misconception that laying more track adds capacity.  (I’m approaching this as transit is either bus, train, or trolley – but using ‘tracks’ is an understood infrastructure that transit runs on, even if buses run on roads.)  Here are some loose, meandering, but interconnected thoughts on how to better understand and differentiate the notion of capacity in relation to cars and public transit:

In transit, you want as many people riding as possible due to the inherent spatial efficiencies of transit in cities where space is inherently in competition (i.e. congestion) vs. cars in order to help with congestion.  Therefore, you should always be concerned about ridership as a measure of utility and because of the external positive multiplier effects that occur from transit ridership.

Lots of ridership means the service is generally a useful and desirable alternative if not primary means of travel.From the operations side, high ridership usually means high demand (in the sense that cities are about maximizing the availability of choice without compromising choice), so that transit agencies can charge somewhere closer to market price.  I’m making an assumption that some subsidy is ok and should always be considered because transit is a public good, can be an agent to help people out of poverty, and can help shape real estate dynamics, but the more that the market is willing to pay for high quality service, all the better to offset a greater portion of operations costs.  For example, an annual oyster card might run you close to $6,000/year, but you also have access to the entirety of London in your wallet without having to own or park a car.

Transit scales with demand cheaply and effectively in a positive feedback loop.  Because (and this is a point I’m going to recite over and over) the capacity is not in the infrastructure.  Whereas highway capacity cannot scale with demand because the capacity IS in the infrastructure, baked into the initial capital costs.  In transit, capacity is part of operations costs.  For highways, tolls are a way to address peaks and valleys in demand (even if nobody likes paying for them because we’ve become accustom to the massive subsidy of free, happy motordom).

The more useful transit is, the more it is used, the more revenue it generates, the better service can be delivered, the better the entire city functions.  With highways, more cars on the road means the road functions less efficiently (one of many reasons why highways are inappropriate within urbanized areas.  I say urbanized, because urbanized areas comes in many forms and highways as well as any heavy infrastructure can touch these areas tangentially.  Tangents good, secants bad.  Transit service goes up with ridership (and ideally technology), whereas with highways service drops with increased traffic.

In this way, we have to ensure the transit design (infrastructure) drives demand and the capacity (operations) can be provided in a way to reinforce that demand.

We know that highways drive demand via induced demand due in part to Braess Paradox and another part to Turner and Duranton’s Fundamental Law of Traffic Congestion.  Braess is more about the psychology of driving, everyone acting in (unenlightened) self-interest will take the assumed fastest route and thus clogging it while the side roads remain the better option (which alludes to Jevons Paradox); whereas Turner and Duranton (and others) discovered the more highways we build, the further we live from our destinations, the more we drive and thus, the more congestion we create (which builds upon Marchetti’s Constant).

Roads are also different in that they externalize operations of individual vehicles to the private sector.  For the most part, the public sector builds the capacity (at the very least thinking they are acting in the public good) and private auto owners have to purchase, maintain, and operate the vehicles themselves.

Whereas in transit, building the lines is a fixed one-time capital cost, but the purchase, operation, and maintenance of the vehicles is still embedded forever into the annual balance sheet.

Building new roads means adding capacity, typically.  The caveat is that all too often the transportation planners are thinking about limited access roads that disconnect local networks and instill incentive to the real estate market to sparsify.  Distance is also a form of disconnection, which is why infrastructure should be planned, design, and invested in to instill the value of proximity within the private sector response. That’s simple physics:  the cost of energy expended in relation to the benefit of the value of the interaction between two parties.  Social and economic exchange done efficiently.

Furthermore, transportation planners all too often think only of the corridor and not of the overall network.  Or, if they’re thinking of the network only as unimodal rather than multi-modal.  Bus planners do bus planning. Bike planners do bike planning. Highway planners do highway planning. And everyone forgets about the city grid. That’s only partially a joke.

The more interconnected the road network, the less travel demand it creates.  That’s a win-win because highly interconnected networks as measured by intersection density and parallel routes are a form of capacity in the form of choice: more people able to get to more locations in more different ways.  Less vehicular demand, but more capacity is how you reduce congestion.

Reticulated grids have more capacity than dendritic because of the funnel points of branching structure of dendritic grids.  You can aggregate all of the lane miles of reticulated grids and the sum is the capacity whereas in branching structures the capacity is only the largest trunk that everybody must funnel through.  Route choice matters.

Whereas the more interconnected a transit or bike network is the more demand it creates.  Networks not corridors.

The capacity of roads is the road network: how many there are, how wide they are, how interconnected they are.  The capacity of transit is not the bus lines or the rail lines.  Let’s call this ‘theoretical capacity’ – what people misconstrue as capacity, the one time capital costs.

The real capacity of transit comes from the amount of vehicles (trains or buses) and the seats within the cars, which are limited by operations costs and revenue.  Recall the internalization of operations from point 4.  The capacity of transit is the amount of cars/seats that you can afford to run given your revenue.

When you treat one time capital costs (and debt) as (virtually) unlimited, but your operational costs for cars, drivers, and frequency of routes is limited, you’re going to have a bad time.  Never French Fry when you should Pizza.  Ask DC metro what happens when you over-extend yourself and in the process sacrifice maintenance and service quality.

If you lay too much track or draw too many bus lines (Coverage over Frequency) given your limited operations revenue you stretch your operations too thin which reduces the impact of one of the primary demand drivers of transit, frequency and reliability, instilling a negative feedback loop on ridership.  Great.  You have a bus line or rail line, but not enough trains or buses to run on it diminishing demand.  Like traffic, ridership is a gas not a liquid.  It is not a zero sum game.  You either attract more or you attract less.

In that way, adding more theoretical capacity undermines your ability to a) add real capacity (greater frequency = more seats moving past more often with less headways, the wait between) and b) have the flexibility to scale with demand that c) you drive with increased frequency; thus, by thinking you’re adding capacity you are actually adding less capacity.  The demand component is critical.  Buying bigger buses or bigger trains without the demand to fill them is still ‘theoretical capacity’ since the time waiting between buses or trains is more of the demand driver than the amount of seats.

I suspect this is partially what’s at play with this graphic:

Light Rail Ridership per Station per Year
Light Rail Ridership per Station per Year

I was looking at annual ridership numbers when I noticed that Light Rail (LRT) ridership jumped every few years.  I was interested to see whether those jumps in ridership coincided with the opening of new stations.  When I normalized for the amount stations open or opened in that year and measured against monthly ridership in December for each year, I got the chart above showing that ridership is dropping as we expand the system into new areas.  Yes, expanding into new areas is a form of capacity if you can maintain frequencies and level of service, but the downward slope suggests a system that is getting less efficient (more capital costs to expand the infrastructure, more operations costs to operate the capacity and maintain the infrastructure, and revenue dependent upon a fickle sales tax allocation) and hopefully not to the point where service cuts are required.

Part of this is that expansion usually means into areas not ‘transit ready,’ lacking the context supportive of transit ridership.  However, that doesn’t explain why Dallas County transit ridership as measured by regular commuters was higher in 1990 when there were no trains than 2015 by raw numbers (39,986 to 36,821) and mode share (4.24% to 2.99%).  The amount of commuters in Dallas county also climbed from 943,000 to 1,231,000 in this time frame.  Dallas County density climbed from 2,105/sq mi to 2,930 over this span as well.

Transit capacity should be designed in a way that creates new demand (i.e. capturing more mode share).  That means delivering maximum level of service where transit is most needed.

Maximizing cars and seats within resources yet still achieving a robust network (while not duplicating services and thus wasting resources) in order to maximize frequency to geographies of demand.  Not only is operations a demand driver (frequency), or layout of the system (network that reaches more places w/o sacrificing frequency), but so is geography (dense clusters of jobs, services, amenities, and/or housing).  Good transit design allocates the most resources to those three demand drivers so that it can also deliver services elsewhere.


  • Cristine

    The challenge of transit and its future is not new and we still have yet to learn from our past.

    Bartholomew’s Reports in the mid-1940s addressed issues, future populations, and preparing for them.The fifth report in the series was on Transit Facilities (1944) and emphasized:
    • Importance of transit facilities,
    • Effect of the auto upon transit facilities,
    • Transit facilities will be even more essential in the future, and
    • Factors considered in this report:
    Operating details that may be adjusted when and where needed: location of routes (major influence), type of service, schedules, and fares.

    Just as the major street plan provided principles, the transit facilities report provided specific principles of a modern transit system:
    • Unified management,
    • Type of service,
    • Area of service,
    • Alignment of routes,
    • Speed, and
    • Headways
    There was a sense of urgency that these principles should be followed in order to provide an efficient, quality of life transit system for the urban community.

    In 1957 – a report: Dallas Metropolitan Area: Thoroughfares, addressed similar issues. With the disappearance of the street car system in favor of an expanded bus system, the report pointed out that efficiency of the expanded bus system was dependent upon an efficient street system. In addition, efficient traffic engineering will aid in reducing congestion and engineering principles will help guide future street designs that achieve efficiency and avoid costly modifications.

    My concern is each street that is given a road diet or restored to two-way may very well be detrimental to the public transit system as a whole. We need more in-depth conversation regarding our streets as a system serving the city as a whole. It seems that decisions are currently being made based on the particular instead of the whole. Streets are too vital to be part of hodge-podge planning.

  • richardlayman

    I don’t think the headline and the introduction makes the right point. You use the phrase “more tracks” when you are not talking about “redundancy” on a current line, but “extending” the system. It’d be clearer to distinguish between the two.

    And yes, by the way, when you extend the system, you aren’t adding more capacity, but using it up, and depending on the nature of the service profile, to do this without impinging on service quality, additional steps adding capacity may need to be done at the core. This by the way is the source of many of the problems being experienced by the WMATA system, as an extension was built (Silver Line) without addressing the issues it created on the existing system.

    Anyway, UMN CTS has a great paper on system expansion and it makes a similar point to yours, that the greatest capture of riders is within the core of the system, not through outlying extensions.

  • richardlayman

    not particularly relevant but you might be interested in my piece on a semi related issue with the DC Metrorail system.