How to make the bus better

The transit workhorse needs more than cheerleading

This is a free preview post of Slow Boring, a daily blog and newsletter about American politics and public policy. If you like what you see, I hope you’ll consider buying a subscription.

Every once in a while someone rediscovers that city buses are a good mode of transportation, either because they’ve gotten frustrated with a local government wasting money on mixed-traffic streetcars, or else — as in this recent Farhad Manjoo column — because they’ve gotten frustrated with endless Silicon Valley hype-cycles about new driving technology.

And this is true. The bus is a very elegant solution to the basic geometrical issue with cars. Forty people who fit very comfortably on a bus would fill several lanes in solo cars. And if they’re standing, even more people can crowd onto a bus at peak hours, while the street literally cannot accommodate any extra cars.

Self-driving car technology would not fix the geometrical problem and indeed might make it worse by encouraging people to drive longer distances. But self-driving bus technology might solve some problems with existing buses. So I believe in the bus and its enduring role.

But the big problem with the American bus is that while a bus can hold a lot of riders, the typical American bus doesn’t hold many riders. And because the buses are running empty, they are very expensive on a per-rider basis, even though it’s not like we’re spending a fortune on buses. In most of the country, transit is marginal, and even better buses wouldn’t make a big difference. But it does matter in the minority of the country where transit isn’t marginal. And especially if everyone gets smart and reads my book, we’ll need to deepen transit availability.


So let’s learn about the bus. My ideas here are so thoroughly influenced by Jarrett Walker and Alon Levy that I am not going to bother to cite them on every particular point. I’ll just say that if you want to understand mass transit, you should read Human Transit and Pedestrian Observations. I’m just trying to convey their basic ideas to a general audience.

Beyond bus stigma

Bus Discourse in the United States is haunted by the menace of “bus stigma,” with a lot of time wasted on discussing it or on coming up with ideas to make the bus sexy by adding Wi-Fi or USB ports.

The truth is that people ride the bus when it makes sense.

  • I was a regular bus commuter when I lived on the 1300 block of Harvard Street and worked on the 1700 block of L Street — taking the 16th Street bus lines downtown was the fastest way to get to work.

  • I was a sporadic bus commuter when I lived on the 1300 block of Florida Avenue and worked on the 1300 block of H Street — I biked in nice weather but rode the bus when it was inclement.

  • Now I live on the 1300 block of Riggs Street. And while I walked to work at Vox, I would routinely take my son downtown on the bus to the Smithsonian museums (pre-Covid, when they were open) or the National Zoo or even just up to the IHOP and indoor playground in Columbia Heights.

One thing these scenarios have in common is that they are all destinations that are popular enough for parking to be scarce and/or expensive. When good transit exists, it’s broadly beneficial and useful to all kinds of people who — for whatever reason — don’t want to drive a car for that particular trip. But the mainstay of transit use is going to be taking people to crowded destinations. If those destinations don’t exist, transit will always be marginal. And if your land-use policies make it impossible for them to exist, then your transit improvement is going to be pushing on a string.

But the other thing these scenarios have in common is that there are frequent bus lines going where I am going. I’ve spent most of my time in D.C. living between 13th and 14th Street, and D.C. maintains strong north/south bus corridors on 16th Street, 14th Street, and 11th Street.

That’s how we get people to ride the bus. First, there has to be someplace to go where driving your car has some downsides. Second, there has to be a frequent bus that goes there.

Designing bus networks for ridership

Most American cities design their bus networks to maximize coverage — they want to make a map that has lines all over it showing that no matter where you live, there’s a bus for you.

The problem here is that for most trips in this town, a car is a very convenient way to get around. So you have lots of buses plying routes where nobody wants to ride the bus. And because the vehicles are looping around everywhere, service isn’t very fast or frequent.

With the exact same vehicles and drivers, you could do this instead.

Now if you live on one of the two dense corridors, you have frequent bus service to downtown. It’s not quite as convenient as driving your car downtown, but you don’t need to pay for parking. Lots of people live in places with no bus service at all, so they have no choice but to drive, but the upside is that they get bigger lawns.

Jarrett Walker, who made these maps, says it’s not his job to tell cities whether they should prioritize ridership or coverage — just that there’s a tradeoff. Typically, cities that go through his planning exercise end up setting a goal that mostly prioritizes ridership but still gives some weight to coverage.

I’m not a consultant, so I’ll just say you should prioritize ridership. But that’s for later. The basic point is that one reason cities tend not to get as much ridership as they could is that the current planning paradigm does not say “allocate your bus resources so as to maximize bus ridership.” A better idea would be to try to get more riders.

“Frequency is freedom”

The Walker slogan at the core of sound bus planning is “frequency is freedom.”

When a bus arrives only once every 30 minutes, then even short trips have a very long time to completion. But worse is the inconvenience of missed trips. If a bus comes every five minutes, you don’t worry about missing the bus. If a missed bus means a 30-minute delay, then you need to go out of your way to get to the bus stop early to avoid missing it. Recall that the bus-riding population is more downscale and more likely to have service-sector shift jobs where lateness can be crippling. Long waits also make transfers unreasonably difficult. Nobody loves transferring, but transfers are the only way to make a rich mass transit network work. But you can’t ask people to ride a bus, get off, and then wait 22 minutes for another bus.

How frequent is frequent enough? They say 10 minutes is about the threshold at which you can forget about the schedule and just show up.

Obviously if you have a high demand on some lines, then bumping the frequency up even more (at least during parts of the day) is great.

But the point is that if you have a promising transit corridor, you should try to deliver service that arrives at least once every 10 minutes during the main portion of the day. Most people still won’t ride the bus! But people who do live on the corridor and are going to another destination on the corridor plausibly will. And a few strong corridors sets you up for complementary land use and change over time as well as further expansions.

Better fewer, but better

The complement to trying to deliver quality, high-frequency bus service on the most promising lines is that your city should probably eliminate some of its lowest-performing lines.

If infinite, free money showers down from the heavens, maybe you don’t need to.

But providing frequent service on the most promising lines is expensive, which is why it’s often not done. So reallocating resources to where you have the highest potential to win is smart. Beyond just eliminating lines that are total duds from a ridership perspective, you often need to make the tough choice of consolidating lines.

In other words, you might have two semi-strong bus corridors, running roughly parallel to each other, each of which gets served every 12 minutes at peak and every 20 minutes off-peak. If you can eliminate one of those lines and create a single line that’s twice as frequent, that will be better. Yes, some people will have to walk further. But everyone will get a more functional transit line. The problem politically is that of course everyone will have a preference as to which line to eliminate, and you need to make some people mad for the sake of a better overall system.

The same is true of bus stops. Of course everyone wants a stop right where they board and right where they disembark. But a bus that stops super-frequently is going to be slow as hell.

Eliminating stops to make the bus run faster has concentrated costs but diffuse benefits, so it’s difficult politically. But a faster bus is good both because it runs faster (which people like), and also because faster buses run more frequently (because they turn around and run the route again).

How close should bus stops be? The best practices drawn from Europe suggest 400 meters (about a quarter of a mile), though some of the cutting-edge research suggests going bigger than that. Levy and Eric Goldwyn came up with 480-meter spacing for a proposed redesign of the network in Brooklyn. The key thing is that most U.S. bus routes stop way more frequently than 400 meters, so there’s a big room for change.

Obviously, abstract math on this needs to be tempered to the fact that you want buses to stop where transfers to rail or to other lines are possible. This is one reason why it’s helpful to redesign networks all at once rather than doing piecemeal reforms. You want to consolidate stops, but you don’t want to eliminate major transfers. And when you are consolidating lines, you mostly want to keep the high-ridership ones in place. But in some cases, it’s possible that your stop spacing on perpendicular routes works better if you make different choices. You need to look at the whole map.

Last but not least, bus ridership potential is largest in the big cities that also have rail transit. And you do need to look at the whole transit system.

Fare integration and free transfers

Standard practice in the United States is often to employ a market segmentation logic, where the bus is “worse” and therefore cheaper to ride than rail options.

The best practice is to divide the whole metro area into fare zones, and then the cost of the trip depends on how many zones you travel through. It doesn’t matter what kind of transit you take or how many transfers you make. A one-zone trip is a one-zone trip, and a three-zone trip is a three-zone trip. Here’s the map for the Munich Fare Union area, which I think is a good point of comparison for big U.S. metro areas.

Greater Munich has about the population of the Atlanta or Miami combined statistical areas and it’s less populous than Dallas or Houston. Obviously, for built environment reasons, those sunbelt metros are not going to be delivering European levels of transit ridership. But these are all places that have built rail transit and ought to learn how to operate an integrated system properly. And there’s no reason at all that Philadelphia, New York, Chicago, D.C., and Boston can’t deliver results comparable to a smaller European metro area if they copy best practices.

The point here is that it’s not fundamentally pro-bus to make the bus cheaper than other forms of transit, because you end up generating a lot of waste.

As Levy says discussing the current transit situation on Long Island, “in the morning rush hour, trains run full to Manhattan and empty outbound and NICE buses, which carry working-class reverse-commuters, are the opposite. Thus, half of each class’s capacity is wasted.”

This is absurd. The way the system ought to work is that reverse-commuters also ride the train (it’s empty after all), and most of the NICE bus routes are oriented perpendicular to the LIRR routes to facilitate transfers rather than replicating them.

Of course, this starts to get into a conversation about how to upgrade America’s sad “commuter rail” efforts into something approximating the S-Bahn, which I’ve written about several times. Back to the bus, though, — the key thing about reforms that maximize ridership given a fixed pool of resources is that getting more riders is the way to get more resources.

Feedback loops — reform, ridership, clout, reform

On a political economy level, the biggest problem with U.S. mass transit policy is it’s always conceptualized as something that someone else is going to ride, which is good because it reduces traffic congestion and now it’s easier for you to drive your car.

This is a conceptual dead-end. Telling planners to focus on maximizing ridership — a bit more like what a private company would do — is good not because transit needs to try to make a profit, but because for transit to succeed, you need a powerful coalition of people who actually use the transit. And with the bus in particular, there are a lot of feedback loops.

  • If a line gets better, more people will ride it, which leads to more farebox revenue, which makes it viable to run buses more frequently.

  • If ridership is high, the political economy of creating dedicated bus lanes gets better, which leads to faster buses and even higher ridership.

  • If you can speed buses up enough on your key routes, then you can deliver good frequency with fewer vehicles, which lets you add more routes and build a stronger overall network.

But the biggest and most important feedback loop is land use.

In a city with good land use policy (i.e., no actually existing American city), a marginal improvement in the quality of bus service ought to induce a marginal increase in density, which generates a marginal increase in bus ridership. And it should be the same on the destination end. At first, a few people start riding transit to Destination A because parking there is a bit annoying. That increase in customers at Destination A makes it economical to redevelop one of the parking lots into an activity center of its own. Now there’s more stuff but less parking! So it makes even more sense to consider the bus.

Now for this to work, regulators can’t be mandating minimum levels of parking, requiring building setbacks, capping building heights, etc. the way they do in every city (yes, even Houston). And I think it’s backward to say you can’t change land use until you’ve changed transportation options. Americans are very familiar with cars and parking, and the market will always ensure that parking exists. But it is an increase in the market price of parking (either in the sense of “you need to pay to park your car at the destination” or in the sense of “you could save a lot of money by not owning a car”) that motivates transit ridership.

No amount of bus-cheerleading is going to change the fact that it’s more convenient to drive your own car. The bus solves a geometry problem about throughput and vehicle storage, and if you try to prevent that problem from occurring via land use regulation, then only the desperately poor will ever ride a bus.

Is everything different now?

A reasonable question to ask is whether or not this is all irrelevant today with the future of the commute in doubt thanks to remote work, etc.

I think the answer is that it matters less in some ways and more in others.

One way it matters less is that traffic jams, for example, should become less of a binding constraint on the growth of the overall American economy, which is good. Using Zoom and Slack, to some extent, is a substitute for improving American transportation.

The flip side of this, though, is that if you are a public official in Greater New York, it’s now more urgent rather than less urgent to make it easier to get around. The fact that people have a greater ability to evade your region’s transportation problems by moving to the suburbs of Raleigh means that you have to become less complacent about actually solving those problems.

Similarly, it ought to be embarrassing to American urban planners that the introduction of Uber and Lyft into the marketplace seems to have reduced rather than increased transit ridership.

After all, once you own a car, it’s usually cheap to drive it somewhere. The way to save money by using transit would be to own fewer cars in the first place, reducing the need to buy one, buy insurance, and obtain parking. But even if there is a decent transit option for your daily commute, in most of the country it’s unlikely that transit can serve all your needs for a typical month. Then Uber, Zipcar, and electric scooters come along and can meet a lot of your edge cases. That makes it much more viable for a single person to own zero cars rather than one, or for a couple to own one car rather than two, while relying on transit for daily commuting trips.

But for that vision to become reality, it has to actually be possible to save money by eschewing car ownership. In my very expensive neighborhood in D.C., the owners of very expensive rowhouses like mine are generally required to maintain off-street parking spaces for the cars we are presumed to own. If it were legal to turn the parking behind our houses into one bedroom ADUs, that would be very lucrative and lots of people would do it. The land would contain more people but fewer cars, and thus more bus riders and all kinds of positive feedback loops would be in effect. Under actual American land use, though, the value of giving up your car is very low, so Uber mostly crowds out bus ridership.

A word on free transit

Twitter and other forms of social media really optimize for radical-sounding takes that can be easily summarized in a snippet, like “mass transit ought to be free.”

This article, by contrast, is a boring technical discussion of stop spacing, route consolidation, land use, and fare integration that is not well-designed to induce other people to hit “share” or “like.”

And I think that it’s true that if you abstract away from all kinds of details and specifics, in some sense transit ought to be free. If your city doesn’t do comprehensive congestion pricing (which no American city does), then you are basically doing your fellow citizens a favor by riding the bus rather than driving your car. Given that, there is no reason to charge you a fare. The cost of operating the bus system is like the photo negative of a congestion charge, so it makes sense.

In real-world budget politics, I think this is less compelling.

It’s possible there is an American city somewhere whose transit could be improved by making the service worse but free. But I don’t think any major cities are like that. Of course you could increase the overall amount of subsidy. But that poses the same question — if you got a bag of magic money, should you use it to improve service or to cut fares? In the vast majority of cases, I think cutting fares is not very compelling.

More to the point, if your typical city got a bag of magic money, I’m not at all sure that spending it on mass transit — as opposed to education or parks or giving money to the poor — would be a very compelling idea.

And that, fundamentally, is why it’s important for transit fans to learn about and advocate for best practices. Most Americans don’t live in places where mass transit is relevant. If cities can do a better job with what they have, they’ll create the circumstances in which they can get more riders and thus the political circumstances to get more resources for transit.