America has better geography for fast trains than France. By Matthew Yglesias

www.slowboring.com

15 - 19 minutes

Population geography is, obviously, an important factor in transportation infrastructure.

Whenever I see something like Alec MacGillis marveling at how a very modest-sized German city enjoys frequent rail service, I always urge American railfans to look up population density figures. Germany has over six times the population density of the Lower 48 — in other words, it’s like Two Billion Americans, which would be a terrible title for a book. In the U.S., Denver has a larger and busier airport than you might expect based on the size of the city. And that’s at least in part because the geography is unusual. No major population centers are close to Denver, but Denver is centrally located to the whole American West, which makes it a great hub for an airline.

Geography doesn’t determine everything, though. France, for example, recently banned short-haul domestic flights in favor of reliance on its strong domestic rail network.

Because while the country has “only” three times the density of the United States, they have a better train network than Germany anyway because they invested with more alacrity in high-speed rail than the Germans did. The geography of France is actually mediocre at best for HSR. What you want ideally is a population that’s not only dense but strung out in a straight line like in Japan or Italy. France has a big city in the middle, and then a bunch of people living in all directions.

If you want to see a place where the geography suggests we should be using fast trains instead of short-haul plane flights, I would look to the northeastern United States. The country as a whole is big and empty, but the Northeast is much denser than the national average and its cities are conveniently laid out in a more-or-less straight line.

The French high-speed rail network is a little hard to describe because the lines become quite complicated near Paris, where it’s centered. The French capital has four different high-speed terminal stations, plus this weird bypass loop that goes around the city to the east. The lack of through-tunnels means that you can’t take a train that starts west or south of Paris, goes into the central city, and then heads to a point north or east of the city. This schematic map depicts it clearly:

The terminal stations are a legacy of the original infrastructure projects. The cross-Paris rail tunnels are currently used for locally-focused RER service, a completely defensible policy choice given the value of that local service but a major headwind for the HSR network.

But more broadly, the dream scenario for trains is to have many destinations, all in a row. The hub-and-spoke nature of the French population distribution just isn’t ideal. Bordeaux is in a different direction from Paris than Lyon. Serving Nice via Marseille is a non-trivial detour relative to a direct route. Cities like Rennes, Nantes, and Strasbourg just aren’t that big, and ideally, those locations would not anchor routes — trains would just stop there on the way to bigger cities.

That this system works at all is a triumph of engineering over geography. If you want to see a country whose geography is really well-suited to high-speed rail, forget France and look at Acelaland — a very wealthy, very densely populated nation of roughly 50 million people living in a series of large and medium-sized metro areas that lie along a straight line.

This is one of the best potential HSR corridors in the whole world. It’s also a place that would really benefit from HSR because several of its major airports are badly congested. That means faster and more frequent trains to serve intra-regional transportation needs would complement air connections to more distant locations that right now are uneconomical to serve because there is so much demand for short-haul air travel around New York.

What’s more, while I don’t think it would ever make sense to try to build a genuinely national American high-speed passenger rail network, this core Northeast Megalopolis service area is so strong that it might make sense to build extensions through Pittsburgh to Chicago and through Richmond down to Atlanta.

France’s top-performing line connects Paris (12.5 million people) to Marseille (1.75 million people) via Lyon (2.3 million) and Avignon (530,000). That’s a distance of about 792 kilometers.

Consider a train built to roughly French specs running south from New York (19 million people) to Raleigh (2.1 million people) via Richmond (1.3 million), Baltimore (2.8 million), Philadelphia (6.2 million), and D.C. (6.3 million). Even if you assume that car-oriented Sunbelt cities like Raleigh and Richmond badly underperform comparably sized French cities, the American corridor is clearly far more promising. D.C. and Philadelphia are just so much larger than Lyon and Marseilles that even if Raleigh only generates the level of ridership you’d expect from a small French town, it’s still a much stronger route than France’s strongest line.

And that’s too pessimistic. Leaving car trips entirely out of the equation, there are something like 30 flights per day between Raleigh and NYC’s airports. If the option of a 3-hour-and-10-minute train journey existed, it would be like the current D.C.-NYC market where there is healthy ridership for both air and rail. There are also 16 Richmond-NYC flights daily. What’s more, both of those cities are served with lower frequency by air to Philadelphia, D.C., and Baltimore. A train would not only compete with those routes on time but would easily beat them on frequency since it’s anchored by the high demand for travel to and from New York. That southern leg is just dramatically superior to France’s best line. And that’s before you consider what’s north of the city.

You can travel from Paris to London (two very large cities) on the Eurostar in about 2 hours and 20 minutes. North of New York is Boston (5 million), which is quite a bit smaller than London (14 million). The only real city between the two capitals is Lille, which is about the size of Providence. On the U.S. side, you have New Haven which, at 870,000, isn’t nothing. But we still look worse.

Here’s where geography saves the day, though. Boston is closer to New York than London is to Paris. The equivalent distance and train time is the ride all the way from Boston to Philadelphia. And Boston + NYC + Philadelphia is more people than London + Paris. Even better, Baltimore and D.C. exist as well and you get those connections “for free.” It wouldn’t necessarily make sense to build the infrastructure to create a 3:15 train between Washington and Boston just for its own sake. But it exists naturally as a result of the other projects, and it would be competitive with air travel at that distance, as would Baltimore. Richmond and Raleigh are much less competitive, but again — you get them for free, and plane-hating weirdos would use the route. I do not recommend building infrastructure for the sake of plane-hating weirdos, but the fact that they get to free-ride on infrastructure that makes sense for other purposes becomes relevant at a later stage of the analysis when larger cities come into play.

It’s worth talking about rationales here. The Euro mindset is that displacing short-haul air travel onto trains is climate policy.

It is true that emissions reductions at the margin are helpful, but I am less impressed by this than they are. There are just way too many useful air routes that are not viably replaced by trains to think that railstitution is going to be our salvation here. That’s especially true if we’re hoping to see economic development in Africa over the next two generations — that’s going to generate tons and tons more air travel demand in the future. Which is just to say the aviation piece of the carbon puzzle really does need to be solved on its own terms with some kind of technological breakthrough. HSR has a lot of useful applications, but reducing total air travel probably isn’t one of them. Climate change is a real problem. There should be a carbon tax to capture the climate externalities of burning jet fuel, and if there were such a tax, it would further tilt demand at the margin toward trains. But the trains are not per se a solution to the climate issue.

So what’s the case for doing this?

One set of reasons relates to economic development. The very strong anchors of New York, Boston, and D.C. mean it makes sense to provide frequent service all along this corridor. That means smaller markets like Baltimore, Wilmington, New Haven, and Providence also get more frequent service than they “deserve” on their own terms. That’s a spillover benefit of running a train on a dense corridor that doesn’t exist in the point-to-point aviation model. By the same token, existing conventional rail corridors like Springfield to New York via Hartford gain value from the HSR spine.

But the other set of reasons is that in this particular part of the country, better trains would complement aviation.

The United States has exactly three “slot-controlled” airports where demand to schedule flights far exceeds airport capacity, so permission to fly is administered as a regulatory matter by the FAA. Those airports are JFK and LaGuardia in New York and DCA in D.C. Now to an extent, we could improve this situation by auctioning the slots rather than administering them as a regulatory matter. But even with improved allocation, you’re still left with the basic reality that more people want to fly to NYC than its current airports can accommodate. There are also four more airports at a slightly lower tier of slot pressure — Chicago, Los Angeles, San Francisco, and Newark.

Again, New York is a huge city. People want to run lots of flights there. So building a train that can handle the demand for travel to NYC (and to a lesser extent D.C.) is good for reasons other than “trains are good.” The number of flights in and out of New York wouldn’t actually fall; they would just end up serving destinations that are further away. By building good trains, we can use our airport capacity for the stuff that airplanes are especially good at. That’s why the whole focus here, ultimately, is on routes that reach New York City.

France also has a train that goes from Paris to Bordeaux (1.2 million) in slightly over two hours, and they are working on an extension to Toulouse (1.3 million) that will take about an hour.

A high-speed train taking the slightly circuitous route from New York to Toronto via Albany (900,000), Syracuse (650,000), Rochester (1 million), Buffalo (1.1 million), and Hamilton (800,000) would take about 3 hours and 20 minutes. There are just a lot of people in the northeastern United States, so easily-overlooked metros like Rochester and Buffalo have almost as many people as the top five French metro areas. New York is bigger than Paris, Greater Toronto is much larger than any non-Paris French city, and little Albany, Syracuse, and Hamilton are basically just along for the ride.

Across the respective cities’ various airports, there are several dozen daily flights between New York and Toronto. So even though this is the kind of distance where fast trains would compete with flights rather than dominate them, it should still support frequent service that displaces some of that air traffic.

And that’s a perfect example of the kind of benefits I’m talking about — you’re decongesting New York City airspace (more slots for flights to Calgary) and creating a transportation windfall for those smaller intermediate cities. You also get the important “for free” benefits of the larger northeast corridor. Are there a ton of people clamoring to ride a train from Baltimore to Syracuse or from Philadelphia to Buffalo or D.C. to Albany? Of course not. But it’s not zero people.

Another interesting spur off the main northeast corridor spine could go from Philadelphia (still 6.2 million people) out to Pittsburgh (2.3 million) and Cleveland (2 million). That’s a bit of a marginal proposition, but you can do the Philly-Cleveland run in just 2.5 hours, which means that with the “for free” connection to New York, you’re looking at a 3:10 connection between New York and Cleveland and 2:30 to Pittsburgh. Again, it’s not just that a lot of people live in New York (though obviously they do) but that decongesting New York City’s airports has a particularly high value.

Cleveland is closer to Chicago than to Philadelphia, and Chicago is a very large city. I don’t know that there’s any particular reason America needs a high-speed rail line going from Cleveland to Chicago via Toledo, but it is right there as a possibility.

And I think there’s a big known-unknown about this sort of line. If you look at European or Japanese high-speed rail ridership just based on city sizes, then New York/Philadelphia/Pittsburgh/Cleveland clearly has a lot of potential compared to many French HSR routes. That said, I think a reasonable person’s intuition is that while residents of northeast corridor cities may behave similarly to people living in European cities, someone living in a much more auto-oriented metro area like Cleveland probably wouldn’t. But given the sizes of the cities involved, the route works even if Cleveland and Pittsburgh underperform relative to their populations.

If you built the line, you would know how much they would underperform.

If they underperform just a little, then we’ve learned that the Rust Belt cities enjoy HSR just fine and the extension to Chicago makes sense. Relatively few people would actually take the entire 4:40 trip all the way from New York, of course, but the point of a train is that people get on and off at each stop. By the same token, if Richmond and Raleigh turn out to do well, then it makes sense to keep going south to Charlotte and Atlanta. Just based on the city sizes, you’d expect a Raleigh/Charlotte/Atlanta line to generate enough ridership to justify the construction, and then you’d get the long-range routes to D.C. and Philadelphia for free. I am, to be clear, deeply skeptical that this would be true in practice. But that’s again just to say that you can start with the more modest plan, get the data, and then extend it further south piecemeal if that seems warranted.

It probably won’t be. But if it was warranted, then you’d eventually find yourself building out routes from the Chicago and Atlanta hubs.

All that being said, I really do want to get away from flights of fancy (Nashville to Jacksonville!) because even though flights of fancy are good for going viral, they make normal people ask questions like “what the fuck is wrong with these railfans?”

That’s why I think the comparison of the northeastern core with French geography is interesting and important. The United States of America as a whole has lots of domestic transportation needs that can’t be met by trains. The United States also has a great domestic aviation system, with multiple large airlines that compete against each other, plus a few scrappy startups. The status quo is mostly pretty good and doesn’t call for huge new construction projects. But the constrained airport capacity in New York City (and D.C.) is a genuine pain point for the system. And while America as a whole doesn’t have great geography for fast trains, New York really does. The northeast corridor geography is so good that Amtrak gets decent ridership even though its trains are really slow and crappy. With modern technology, you could go from D.C. to Boston in about the time it currently takes to get from D.C. to New York.

That route well done, and perhaps a few other NYC-centered routes, would attract riders, displace air travelers, and free up capacity for planes to fly further away. And it would stitch together a region of maybe 60 million people that, while only a geographically small slice of the United States, is home to a very large share of its population and economic output.