Does transit infrastructure cause ridership?

Does building a new transit line trigger ridership?  Does it even make sense to talk about the ridership of a piece of transit infrastructure?  

If you say yes, you're expressing an infrastructurist world-view that is common in transit investment discussions.  The right answer to the above questions, of course, is "No, but:

  • Infrastructure permits the operation of some kind of useful transit service, which consists of vehicles running with a certain speed, frequency, reliabilty, civility and a few other variables.
  • That service triggers ridership."

To the infrastructurist, this little term — "service" — is a mere pebble in a great torrent of causation that flows from infrastructure to ridership.  By contrast, service planners, and most transit riders that I've ever met, insist that service is the whole point of the infrastructure.

If you read the literature of infrastructure analysis, you  encounter the infrastructurist world view all the time, mostly in ways that's unconscious on the authors' part but still a source of confusion.  This afternoon I was browsing TCRP 167, "Making Effective Fixed-Guideway Transit Investments: Indicators of Success", which includes some really useful explorations of land use factors affecting the success of transit lines.  But when they talked about infrastructure features as causes of ridership, the report routinely delivered weirdness like this:

The percentage of the project’s alignment that is at grade proved to be a negative indicator of project-level ridership. At-grade projects may be more prevalent in places that are lower in density, while transit is more likely to be grade-separated in places with higher density or land value. Thus, this indicator may be reflective of density. It may also be true that at-grade systems are slower than grade-separated systems. At-grade status may reflect a bundle of operational characteristics such as speed, frequency, and reliability, although the analysis did not find that these factors individually had a statistically significant effect on ridership.  [TCRP 167, 1-17]

This careful talk about how a correlation "may" reflect density or "operational features" sounds vague and speculative when it's actually very easy to establish.  There is no shortage of evidence that:

  • High density reliably triggers ridership.
  • Areas of high density are less likely to have available surface rights of way.
  • Therefore, highest ridership segments tend to be grade-separated.

So this is a case where "A correlates with B" does not mean "A causes B" or "B causes A".  It means "A and B are both results of common cause C".  It's important to know that, because it means you won't get B simply by doing A, which is the way that claims of correlation are usually misunderstood by the media and general public.

Later in the paragraph, the authors again describe the obvious as a mystery:

At grade status may reflect a bundle of operational characteristics such as speed, frequency, and reliability …  

Yes, it certainly may, but rather than lumping all the at-grade rail projects together, they could have observed whether each one actually does.  

… although the analysis did not find that these factors [speed, frequency, and reliability] individually had a statistically significant effect on ridership

While this dataset of new infrastructure projects is too small and noisy to capture the relationship of speed, frequency, and reliability to ridership, the vastly larger dataset of the experience of  transit service knows these factors to be overwhelming.  What's more, we can describe the mechanism of the relationship, instead of just observing correlations:  Speed, frequency, and reliability are the main measures of whether you reach your destination on time.  Given this, the burden of proof should certainly be on those who suggest that ridership is possibly unrelated to whether a service is useful for that purpose.

Note the word choice:  To the infrastructurist, speed, frequency and reliability are dismissed as operational, whereas I would call them fundamental.   To the transit customer who wants to get where she's going, these "operational" variables are the ones that determine whether, or when, she'll get there.  It doesn't matter whether the line is at-grade or underground; it matters whether the service achieves a certain speed and reliability, and those design features are one small element in what determines that.  

I deliberately chose a TCRP example because the authors of specific passages are not identified, and I have no interest in picking on any particular author.  Rather, my point is that infrastructurism so pervasive; I hear it all the time in discussions of transit projects.  

I wonder, also, if infrastructurism is a motorist's error: In the world of roads, the infrastructure really is the cause of most of the outcomes; if you come from that world it's easy to miss how profoundly different transit is in this respect, and how different the mode of analysis must be to address transit fairly.

Whenever you hear someone talk about the ridership of a piece of infrastructure, remember: Transit infrastructure can't get people to their destinations.  Only transit service can.  So study the service, not just the infrastructure!




14 Responses to Does transit infrastructure cause ridership?

  1. Rob August 29, 2014 at 9:43 pm #

    It seems to me there are two types of transit enthusiasts – those who use transit to get around, and those who see transit as an enhancement to real estate. Both like the transit-friendly (which is really pedestrian-friendly) effect of transit infrastructure on surrounding development, but the people who rely on transit as a transportation system are more likely to understand that service is what counts for ridership.

  2. Reality Check August 30, 2014 at 3:00 am #

    What’s more, instead of the infrastructure dog wagging (or limiting) the service tail … the service plan should precede and drive the infrastructure. In order to intelligently right-size your infrastructure, you’ve got to have a pretty good idea of the service plan requirements/goals.
    In the case of CA HSR and/or Caltrain infrastructure design, we have extremely overbuilt and over-costly-seeming infrastructure being proposed (SF Transbay, SJ Diridon Pan Galactic, Millbrae/SFO BART/Caltrain, etc.) completely divorced of concrete service plan goals. So once the infrastructure guys have done all their real-estate acquisition, excavated all their holes and trenches, put all their rebar and concrete down, laid tracks and built platforms and canopies, etc. you run into massively slap-your-forehead stupid over and under-built infrastructure when finally working out the actual SERVICE plan of how and where and at what speed and volume to slot-in all the various trains to meet actual, realistic desirable/realistic ridership demands.

  3. duke the lost engine August 30, 2014 at 3:05 am #

    I agree with everything here, but would partially defend the infrastructurist view in decision making.
    Infrastructure investment decisions have impacts lasting many decades, during which service levels may be revised several times. Service levels aren’t so much a pebble as something that current decision makers just don’t have much control over (aside from the first few years of the infrastructure’s life).
    so it makes some sense to focus on the long term relationship between infrastructure type and mobility. (being careful about causation hopefully!)

  4. Eric August 31, 2014 at 5:10 am #

    “It seems to me there are two types of transit enthusiasts – those who use transit to get around, and those who see transit as an enhancement to real estate.”
    But why would transit enhance the value of real estate, if it’s not useful for getting around? Because it looks pretty? If so, the enthusiasts are wrong for pushing such an expensive form of prettiness to receive public funding. Because the people who buy the real estate will incorrectly think that the transit is useful for getting around? If so, the enthusiasts are dishonest for lying to the buyers.

  5. Simon August 31, 2014 at 2:40 pm #

    Great post.
    But how do you convince the average voter of this?

  6. Eric O August 31, 2014 at 8:39 pm #

    My experience has been that the Infrastructurists, the transit line planners, don’t tend to ever make it into the room where the public is present to discuss service and operations matters. The former are future oriented; the latter crowd is typically focused on immediately addressing the problems of existing services. These folks are therefore present-oriented, and their discussions tend to be driven by their personal experiences using the present services. Usually, these two crowds tend to completely miss each other, like ships passing in the night,… maybe because the agency has two very different fronts of engagement in regards to these.
    When the Infrastructurist types get before the public, they are typically there to discuss the project as infrastructure planning – almost exclusively. The people that show up at their meetings are developer types, transit advocates and impacted property owners who tend to be there to discuss project specific matters (such as alignment alternates, anticipated traffic and construction impacts and so on). The people who tend to be the Infrastructurists thus don’t ever get to sense the public’s high interest (and often frustration, as the case might be) in the actual service outcomes. It tends to be a very delimited discussion that goes on in the project planning phases.

  7. valar84 August 31, 2014 at 10:10 pm #

    I think infrastructure is important for offering service potential. If you only have a regular street to run transit on, a local bus line has a relatively low capacity, a regular bus every 5 minutes is just around 700 people per hour per direction, which is less than a single road lane used exclusively by cars. Worse, as ridership increases, service speed will progressively fall as buses will have to stop more and more often (therein lies the dilemma of local bus service: fast and infrequent, or slow and frequent, but never fast and frequent).
    A rail line however, even an at-grade one, has much higher capacity as vehicles are larger and can be put in trains, stations with pre-payment allow speed to be protected even as ridership increases.
    The point of focusing on infrastructure is essentially a land use approach. It’s much better to build the infrastructure as soon as possible, before an area gets too built-up and land prices get much higher (making transit investments much more expensive). The presence of the infrastructure draws transit-oriented development, for increased ridership often depends most on attracted developments, not on dragging people away from cars, but by getting them to move in transit-friendly locations and concentrating developments there. Of course, zoning has to allow these developments.
    But if you have no infrastructure and you can only run local buses, their low capacity and low speed will soon choke transit-oriented development. People who can afford cars will abandon them and switch to cars to save time and get more comfort while traveling, and once they start taking cars everywhere, they will require parking and wider streets and just worsen the situation.
    As good as it is to try to make transit service better in car-oriented developments, ultimately, we’re trying to force a square peg in a round hole. These developments have not been built for transit service, they were planned for cars with no regards to the conditions to make transit work. Transit is just inefficient for them. On the other hand, if you look at really good TODs like streetcar suburbs, Japanese suburban rail stations, German S-bahn suburbs and the Rosslyn-Ballston corridor, generally the rail lines were built BEFORE the density and ridership potential justified them. The infrastructure came first, development came later, a development centered around the infrastructure. This is how you get good transit ridership.
    I’ve heard something interesting recently “You don’t justify building a new bridge by counting the number of people who are swimming across the river”. Transit investments must be forward-thinking, pro-active rather than reactive, and be tightly linked to development policies. As long as we keep planning transit as an afterthought, no matter how good thinking supports that planning, we will keep choking in cars with transit mode share in single digits (except in earlier TOD developments).

  8. Ernie September 1, 2014 at 7:53 am #

    I tend to think of this subject with the example you gave in Toronto: The difference between the wrapper and the food. Infrastructuralists are visual type of people and their concerns are with things that are clearly visible: The subways, the BRTs, the condos that are built as a result of subway construction. Hence they are concerned with the marketing and the packaging of infrastructure projects. The freedom, the “food” that is often invisible because it is not easy to visualize. Hence the LRT/Subway debates which really don’t matter as much as frequency, service times in transit.
    I think that Jarrett is right in demanding frequency mapping because it visualizes the importance of operations, and helps city planners create infrastructure that complements the current grid and demand. Infrastructuralists would also better understand if an expensive infrastructure project is warranted if operationally it doesn’t make sense or does not complement current services.
    Please do not eat the wrapper because you can’t digest it anyway.

  9. The Overhead Wire September 2, 2014 at 4:48 pm #

    I wonder how much is driven by funding rather than “infrastructurism” I’ve had many project planners tell me that capital funding is easy, operational funding is not. So what would happen if the Feds only funded operations and had a service level and access quotient attached to the money? (Not saying feds should necessarily fund operations but just a thought exercise)

  10. Bol Werk September 3, 2014 at 7:52 am #

    I think you’re exaggerating this “infrastructuralist” attitude, or at least exaggerating how much infrastructure boosters ignore service. There is nothing contradictory about being concerned about infrastructure and being concerned about service frequency, and there is no way realistic way around the fact that adequate service often requires improvements (investing in infrastructure).
    You can make a much more compelling argument against two other groups, who in a sense represent opposite extremes from each other:
    (1) people who don’t care about infrastructure at all. BRT boosters fall into this trap a lot. They support BRT for because it’s oh-so-cheap, while ignoring that it can easily be more expensive than an equivalent rail implementation in the long run because future operating costs exceed present savings in capital expense.
    (Disclaimer: not saying BRT is always bad. But there are a lot of people who think everything should be BRT because the drop-in aspect appeals to them.)
    (2) anyone/anything inflating costs so far beyond the rest of the world. USA costs for new infrastructure are often much higher than other wealthy countries, which can make it hard to justify some expenditures we should be able to afford. #1 may be made up of people reacting to this.

  11. Christopher MacKechnie September 3, 2014 at 10:37 am #

    I believe there to be several promoters of the infrastructuralist attitude, at least in the United States –
    1) Much greater availability of capital funding versus operating funding – previously discussed. The high availability of capital funding enables transit infrastructure to be built in areas that cannot justify the service levels required for good transit even if operating funding was available.
    2) Infrastructure will be there for 50 – 100 or more years. Once you build it it will almost never be taken out. Service can always be ramped up later. Why not get your foot in the door before winds change and your city gets taken over by Republicans?
    3) Local funding matches for new rapid transit lines are often paid for by sales or other similar taxes that face voter approval. Thus, in order to win approval your line must look “nice” to people who will never actually take it. With farebox recoveries so low and the vast majority of operating funding being paid for by the average taxpayer, I have become more sympathetic to this argument over the years. If passengers paid 100% of the operating cost of the transit system then I would not care what the general public thought about it.

  12. Simon September 3, 2014 at 3:51 pm #

    On point 3 above, subsidies aren’t universally high. More effort needs to be expended to reduce subsidies, in many cases higher fares are at least part of the mix. Not to mention unsubsidised driving.

  13. Andre Lot September 3, 2014 at 7:51 pm #

    @Reality Check: your critique is unwarranted regarding CASHSR. The project has been designed exactly as you said it should – with service specifications in mind, including maximum travel time between SF and LA, general route comprising cities to be served etc.

  14. ajedrez September 7, 2014 at 1:29 pm #

    @valar84: Frequency and speed aren’t necessarily mutually exclusive. For instance, as ridership and frequency increase, you are able to justify more limited-stop and express variants of your route. If your bus service runs every half hour, and half the runs are express, local riders get service every hour. By contrast, if your bus service runs every 5 minutes, and half the runs are express, local riders get service every 10 minutes. (Meanwhile, express riders have frequent and relatively fast service)