Matthew Yglesias thinks more carefully about transit than any other US pundit I’ve encountered at his level of celebrity, so his takes are often useful. Today, he has a very broad one, called “American transit agencies should prioritize ridership over other goals.” He begins:
Why does the United States struggle to create cost-effective rail infrastructure? Why do non-NYC American cities have such a hard time attracting mass transit ridership?
On one level, these are deep and complicated questions. But one thing I want to say to the growing community of people who are interested in them is that American agencies don’t deliver on these goals because this country’s high-level governance constructs don’t say they should deliver on them. Of course you won’t find a grant guideline that specifically says “don’t focus on delivering high ridership at an effective cost.” But wasting money is really, really easy when nobody is specifically telling you not to.
His piece spans a huge range of issues, from rail construction methods to bus service planning. I agree with him about rail: We can do it cheaper by making stations more functional and less palatial, as long as we plan enough capacity. On service planning, though, I think he oversimplifies in a way that lots of well-intentioned urban policy wonks oversimplify, so it’s worth unpacking a little. There are two issues with saying “just prioritize ridership.”
- Ridership is not very predictable. That’s why, in our work, we prioritize access to opportunity instead. Actual ridership is affected by lots of unpredictable external events (pandemics, economic conditions, etc), but access to opportunity is the constant thing that transit provides that is the foundation of ridership. Access is also important for a bunch of other policy reasons.
- Low-ridership service is sometimes justified by policy goals that matter to people, including, in some cases, racial and social equity.
Yglesias cites my work on the ridership-coverage tradeoff: In service planning, a ridership-maximizing network doesn’t go everywhere and serve everyone, because it goes where the most riders are and runs fast and direct enough to be useful to them. So a ridership goal is in conflict with goals variously described as “leave nobody behind” or “meet the needs of low income people (wherever they are)” or “we pay taxes too so we should get some service.” All those impulses lead to predictably low-ridership service, which I call coverage service. I explain why this conflict is unavoidable here.
You will never hear me say, as Yglesias does, that ridership should be the only goal of a transit service, and not just because I’m a consultant who facilitates conversations on the topic. I won’t say that because the decision is genuinely hard, and there are some good policy reasons for coverage services.
The suburbanization of poverty in many cities has increased the number of low-income people and people of color living in suburban land use patterns that are just inimical to public transit. Those areas have fast roads that are unsafe to cross, no sidewalks, disconnected street patterns that obstruct walks to the stop, road patterns that require buses to make crazy loops, etc. A strictly ridership-based approach would not go to a lot of those places, but will put lots of service in dense inner cities that happen to be increasingly gentrified. The result can be something that is measurably inequitable by both race and income. In other words, sometimes, in some common geographies, there’s a ridership vs equity tradeoff. (We have some unpublished work on this for a major US transit agency that we hope to release soon.)
So when Yglesias says …
Now again, I’m happy to concede that across the entire possibility space, you could imagine a situation in which one route maximizes ridership but a slightly different version maximizes economic development goals or equity goals or environmental goals. But those divergences would in practice be either pretty rare or pretty small.
No, this is not an imaginary situation, and the divergences aren’t always small. In our recent work for Portland’s TriMet, the agency articulated twin goals of ridership and equity, which led to a network that provides low-ridership coverage, but only only in low-income and minority areas. Obviously, low-income and minority people generate ridership all over the network, so ridership and equity goals overlap more than they differ, but they still do differ significantly.
There’s a long-term, high-altitude view where Yglesias is right. High-ridership services tend to create positive feedback loops with urban development that encourage even more ridership. Smarter development could also reduce the need for coverage services over time.
But that’s not the altitude and timescale where most transit decision-making gets done, especially in service planning. Those decisions are made by local elected officials or their appointees. The problem is not just that people are yelling at them to defend their bus stop, although they are. It’s that on a policy level, it’s just not always true that the high ridership network is the high-equity network. That means that hard decisions have to happen. I’m there to help boards reach those decisions, not give them the answer.
I notice that the example you cite regards the deployment of extant fungible resources (viz. buses) during operations, whereas all of Matt Yglesias’ examples regard the project-level construction of transportation infrastructure.
That is, when choosing how frequently to run routes (or whether to run them at all), I think you are certainly correct that there is a ridership-equity tradeoff. Likewise, it probably doesn’t make sense for all transit investment to occur in New York, where it has the maximal ridership bang for each buck. But once political processes that a transit infrastructure should be built along a certain general pathway, are you aware of a circumstance in which ridership and do not align in constructing that infrastructure? Or does Yglesias’ conjecture remain viable in that limited domain?
Not sure quite what you’re asking. Major transit infrastructure in the US often has non-ridership features, such as stations with no ridership potential created for political reasons.
That’s exactly the sort of example I was looking for; thank you!
I’m not sure it’s true that adding transit in New York has the most bang per buck? New York costs are crazy high and most of the city already has good service by american standards, so the marginal benefits of additional transit are limited (although there still are a lot of good projects there, especially if they ever manage to control costs).
Have you used the trolley problem, particularly the meme, to communicate the coverage vs ridership dilemma? Seems perfectly suited for this different problem. Sacrifice 5 people to serve one person – less frequent service means fewer riders, but honoring the lifeline service.
The trolley problem? Metaphors involving violent death aren’t super helpful for encouraging clear thinking about tradeoffs, in my experience! They tend to polarize people.
“The suburbanization of poverty in many cities has increased the number of low-income people and people of color living in suburban land use patterns that are just inimical to public transit. Those areas have fast roads that are unsafe to cross, no sidewalks, disconnected street patterns that obstruct walks to the stop, road patterns that require buses to make crazy loops, etc.”
The problem is, a bus is either being ridden, or it’s not. If such a bus is being ridden, it’s helping on “equity”, but it’s also justified on a ridership basis, so we don’t really need the “equity” justification. If the bus isn’t being ridden, even if there’s a lot of people of color who leave near it, if they’re not riding it, it’s not helping them, therefore, it’s not actually furthering “equity”.
I guess if really wanted to do the race-equity thing objectively, you’d do service planning based on a ridership model where each riders gets some weight according to their race (e.g. white=1, asian=1.2, hispanic=1.6, black=2?). But, I would personally find any such planning model repugnant, and I’m sure many others would too. It would essentially be saying that a person is worth more or less based on their skin color, plus whatever numbers you set these weights to be would always be fundamentally arbitrary and subjective.
I think the best option is to for the transit system to simply say a rider is a rider and allocate service in a race-blind way. Or, if you’re going to look at demographics, do so based on income level (which is already correlated with race), rather than race directly. At the end of the day, a person’s “need” is really based on whether they have the money to afford car payments, not what color their skin is.
If you’re going to make racial-equity investments in certain neighborhoods, it is better to do so by fixing the neighborhoods themselves. For example, adding missing sidewalks so people don’t have to wait for an hourly bus to go half a mile, just to get groceries, is something that should be a total no-brainer. And, using racial equity as a justification to get the broader city to pay for those sidewalks is not unreasonable.
I disagree with the idea that a bus is either ridden or its not ridden. It’s a sliding scale. We have routes that are low ridership compared to other routes–but they are still being ridden. We could get more ridership by putting those service hours elsewhere, but it would remove service from a group of people who are definitely riding–and relying on–that service. Obviously a bus with zero riders ever isn’t serving an equity goal and can be removed, but few routes stick around that serve zero riders. It’s always a question of how few makes it no longer worth it.
I agree with you that fixing the neighborhoods would often be a more effective solution. Unfortunately, it seems those who make decisions about transit and those who have the power to fix neighborhoods are rarely the same people, so the transit system is left to make service decisions with little if any power to adjust the streetscape. At least, that’s how it works in my neck of the woods.
I don’t think there’s much support for reducing equity to ridership. Equity is often about access to opportunity, and that access has value separate from whether you use it.
One thing that irks me about transit discussions is that too many bus agencies see themselves as an island. Probably this is due to being set up that way, but I also don’t see too many people pushing to change it. A city’s transit department shouldn’t be focused on only the bus (and/or rail) system, they should be focused on getting people where they need to go when they need to get there.
Thus, they should be spending time and money on improving walking and cycling too. In so many places I’ve lived in the U.S. that have bus systems, the transit department focused only on the buses. There are huge opportunities to provide “transit” to people where all they need is their own propulsion to get where they’re going — and for many of those people, it would be faster than the bus if they rode a bike! Not to mention that walking and cycling are available 24/7, at the exact moment you want to leave home, instead of having to be on the limited bus schedule. This is probably one of the biggest tragedies in US transportation planning. It’s 95% about the car, 4% about the bus, and 1% about walking and cycling. So then we end up with terrible bus service and the inability to walk or bike anywhere safely or comfortably.
This sounds right, and yet all agencies need to focus. Just because two issues are connected doesn’t mean that the same agency should be responsible for both of them. Running buses and trains is hard, and we need to allow there to be a space where people focus on that without distraction.
The island is not only about not including walking and cycling, it is about other transit agencies as well. It’s depressing that a transit agency in one of the areas that have different transit agencies can not only not consider the impact on connections with other agencies in its planning processes, but actively work to discourage or disrupt these connections out of a fear that the other agency is “stealing” their passengers.
While some work has been done to improve cross-agency fare collection, cross-agency service level coordination is still a huge problem in the San Francisco Bay area and especially the Los Angeles area. Not to mention the fact that seeing so many different colored buses in the same area (UCLA for example) is confusing enough to deter new people from trying transit.
This isn’t just a US issue. I’m looking at a map from approximately 2010 showing bus services operated by the government agency in Sydney, Australia. No private sector bus routes are shown, suggesting that there is absolutely no bus service available outside of about a 10 km radius around Sydney’s central business district. And it’s only been in the past few years that, for example, Transport for Greater Manchester (UK) has published maps showing all bus routes operated by all the various private companies at the same time. The maps are very poorly drawn, but it is a start.
I think Christoph Spieler’s views would line up with Yglesias’ at the higher altitude service / infrastructure planning level. Lots of fancy rail systems in the USA (suburban rail or downtown streetcars) that have such poor service levels and/ or such limited extent that they are unlikely to ever see high ridership (they fail Jarrett’s Transit Ridership Recipe on most or all ingredients — they just aren’t useful enough for enough people). Add in high costs, poor urban design, bad parking policy and you’ve got a lot of waste.
On Strong Towns there’s an article about the rail system in Albuquerque / Santa Fe. Daily ridership = 1,800. Track length = 96 miles. Capital costs around $375 million, or over $200,000 per weekday rider!! Only 24 trips per weekday.
I think there’s selection bias here where you see a lot of the projects that do have these tradeoffs, but most projects that fail to prioritize ridership don’t. Matt gives the example of the California HSR project prioritizing both rural jobs creation and access between small towns in California instead of optimizing the main LA-SF line. You can technically make an argument that bakersfield deserves access too for some kind of equity argument, but I think at that level it’s obviously absurd – until you have an effective direct HSR line running between LA and SF, you shouldn’t be investing anything at all in the secondary goals.
It reminds me of this: https://www.thetransportpolitic.com/2016/04/22/which-riders-matter/. I can see both sides of this. There are various metrics that can be used to grade a project, such as cost per rider-time saved. But other factors come into play. There is no reason to assume that such a metric is the only thing you should look at.
My biggest complaint is that so often American projects seem to be ignoring every conceivable consideration, other than distance. It is easy to argue that for an “everywhere to everywhere” system, but sometimes the trains go nowhere, and as a result, ridership is abysmal. Hard to argue you are serving lots of low income people when you aren’t serving many people at all. Dallas, Denver, Sacramento, you name it. I suppose you could argue it is cheaper that way, but per rider? Probably not. But now we are back to metrics.
There are patterns that emerge that are known to be a bad idea, like running in the freeway median to distant suburbs (https://media4.manhattan-institute.org/sites/default/files/economics-of-urban-light-rail-CH.pdf). Another is distance over urban stop spacing. BART started the experiment over 50 years ago, and if it fails there, it will definitely fail in smaller cities (get ready, Seattle, you are next). There are no metrics to support this sort of project, other than it looks spiffy. It is a freeway mindset — we can imagine taking a train that far. But the reality is that mass transit doesn’t work that way. Serve the urban core with trains first (if at all) before you extend outward. Focus on quality not quantity. Look at the entire network (buses and trains) not just the trains. Little of that is being done in various projects around the country, and we are paying the price.
I think the ridership/coverage situation is a lot more common with buses instead of trains. It may be fine to only carry a few riders on the bus, especially if they are mostly low income riders. But if you spend billions on a new rail system and it only carries a few riders, you are in trouble. Chances are, the train won’t run frequently. The larger (and more expensive) the project, the more important it is to get a sufficient number of riders. Likewise with an “everywhere to everywhere” model. Are you really planning on having rail everywhere in a midsize city?
I think this is where Yglesias has a point. Take any major rail project in the United States and focusing on ridership (per dollar) and you aren’t going to find many examples where it fails. The few times where I could see it failing is when there is some place that either has a significant number of low income riders, or potential future riders. A station may be too expensive. But that seems easy to solve with a simple guideline when building lines. For example, if there is a very big distance between stations, we should make sure that there is no future potential for ridership there. In some cases (like crossing a lake) there isn’t. But in other cases, it is just bad planning.
I agree that focusing on just ridership is not a good idea. But it should be one of the biggest priorities.
“Take any major rail project in the United States and focusing on ridership (per dollar) and you aren’t going to find many examples where it fails.”
I disagree entirely. There are massive failures of two kinds in US rail building.
1 – the infrastructure is too expensive per kilometre or per station. Alon Levy argues the point well at their blog pedestrianobservations.
2 – the US builds a lot of questionable rail projects, regardless of their cost relative to other projects or relative to rail in other countries. In Trains, Buses, People, Christof Spieler has many examples of rail projects that serve hardly anyone, because they don’t have stations near many homes or destinations. Some really bad examples:
– Albquerqurque Rail Runner, 15 stations and 1,800 weekday riders
– Nashville Music City Star, 7 stations, 57,000 ANNUAL riders (270 average weekday riders?!)
Add in the streetcars in downtowns, plus questionable expansions in bigger cities and there are many places where rail was chosen for flashiness, but sent into corridors where there just weren’t many people.
I agree that equity and ridership goals don’t always align. However, a significant aspect of their alignment was left out in the post. Ridership increases can lead to more political attention towards transit, which in turn leads to better service and more ridership. This virtuous cycle is in contrast to the “transit death spiral.” When more people use transit, it becomes more relevant to society as a whole.
Transit disproportionately benefits equity deserving groups. By increasing ridership overall, we can improve transportation outcomes for these groups in the long-term. Overall, equity and ridership are likely to align more in the long-term than in the short-term.
I think the concept of productivity versus coverage is anachronistic. It is sustainable versus unsustainable, and too often transit agencies pick unsustainability chasing urban sprawl to new employment centers and sprawling residential neighborhoods. Certainly, many poor people are being pushed out to the old suburbs, but the sheer density of low-income bus riders in those old suburbs would justify high productivity, sustainable service there. I think it is rare for cities to have small pockets of low-income bus riders all over the suburbs in such a manner that providing transit to them all would be unsustainable. When you do provide high productivity service to a low-income, old suburb, it would attract more bus riders to move there and drive up transit use, creating a sustainable suburban service. Unfortunately, many planners get cold feet when explaining to elected officials from affluent neighborhoods why you don’t believe providing low productivity bus service for their maids is sustainable as they keep moving farther and farther out.
Ed I am not sure where you live, but in the US it is extremely common to find “small pockets of low income bus riders all over the suburbs in such a matter that providing transit to them all would be unsustainable.” I design bus networks for a living and I encounter these places all the time.
But I don’t use “sustainable”, not just because of the confusion between environmental sustainability and financial sustainability, but because transit isn’t a business, so what counts as financially sustainable depends on what level of subside government wants to provide. The productivity-coverage tradeoff gets to the underlying geometry in a way that bypasses that factor.