Some thoughts on high-speed rail – part 2: Hubs and Spokes

“The spatial impacts of the new lines will be complex. They will favour the large central cities they connect, especially their urban cores, and this may threaten the position of more peripheral cities.” [Hall, 2009]

“[T]he wider economic benefits of high-speed rail are difficult to detect, as they are swamped by external factors”, but are likely to be larger in more central locations than more peripheral locations.”[Preston and Wall, 2008]

As used here, a hub is a center of activity, from which multiple (at least three) spokes (links connecting the hub with other locations) emanate. On a network with a tree structure, the primary hub is the point from which the maximum number of spokes emerge. There may be secondary and tertiary hubs on the network as well.

The proposed US system (see Figure) , such as it is, has no well-thought out national architecture. There were a number of independent proposals that have been drawn on a single map.

USA-FRA-thumb-400x239-38918 USA-ARRA-thumb-400x308-38915 USA_HSRA_Phasing_Map-thumb-400x259-38912 USA-HSR_Corridor_Route-thumb-400x300-38921 USA-PIRG-thumb-400x283-38924 Interstates-thumb-400x250-38906 Florida-thumb-400x343-38837 California-thumb-400x453-38930 Midwest-thumb-400x333-38933 NewYork-thumb-400x307-38909

  • The existing Northeast corridor, the only US claim to HSR, such as it is, is part of the national “plan”, though it received the least funding from the American Recovery and Reinvestment Act of 2009 (see the top right on ) . The Northeast has the most developed network with semi-high-speed rail (Acela) running from Boston through New York to Philadelphia, Baltimore, and Washington. This could be described as a New York Hub (though it has not been pitched as such), with current non-high speed lines from New York emanating in particular to Albany and then to Rochester and Buffalo or to Montreal, and spurs from New Haven to Burlington, from Philadelphia to Harrisburg and Pittsburgh, from Washington south to Richmond and Raleigh, and from Boston to Portland and Brunswick, Maine, all of which have been proposed for upgrade to high speed.
  • The proposed California Corridor, is based on a mainline that runs from San Francisco, through California’s Central Valley to Los Angeles, with extensions to Sacramento and San Diego. The long-term vision of the national program has a line from Las Vegas to Los Angeles. With all of the commuter rail already in the Los Angeles region, the network could more accurately be described as the Los Angeles Hub. Even the Sacramento line is more oriented to Los Angeles than San Francisco, despite the distance.
  • The most coherent of the new proposals is the Chicago Hub, which as its name suggests, hubs traffic from other Midwestern cities into Chicago. This proposal has achieved agreement from all of the regional governors, and with a Chicago-based administration in the White House, not surprisingly received a large share of the recent federal allocations ($2.6 Billion).
  • The proposed Florida High-speed rail system runs from Miami though West Broward, West Palm Beach, to Orlando, Lakeland, terminating in Tampa with about 10 stations planned. Proposed additional extensions connecting Fort Myers, Jacksonville, and Tallahassee and Pensacola have also been drawn on maps, but these are farther into the future. This could be described as an Orlando Hub. Though Miami is a larger metropolitan area than Orlando, the branching structure is naturally geographically based in Orlando due to it centrality on the Florida peninsula, as well as it central location vis-a-vis tourist traffic. Tourist traffic is important to this line, as stops at Disney and Port Canaveral have been included. It is anticipated the line will carry 2 million travelers yearly ( 5500 per day on 12-18 round trips), and is 324 miles in length in total. With 10 stations, there is an average of 32 miles between stations, which will bear nuisance costs, and 10 station areas, which will see accessibility benefits. The line is anticipated to run along the I-4 and I-95 corridors for significant stretches, so those areas already see some accessibility benefits (at on-ramps and off-ramps) and nuisance costs (between interchanges).
  • The Northwest region, or Seattle Hub connects Vancouver, Canada with Salem, Oregon.
  • The South-Central region, once dubbed the Texas Triangle, and now the Texas T-Bone, may be described as a Dallas Hub, connecting San Antonio, Austin, Houston, New Orleans, Oklahoma City, Little Rock, and Memphis, among others.
  • The Southeast region is probably best described as an Atlanta Hub, as Atlanta is the key interchange in the region (hubbing traffic from Savannah, Jacksonville, Birmingham, Chattanooga, Nashville, Charlotte, and Raleigh), and the largest metropolitan area. There is also a line from Raleigh through Columbia to Savannah, bypassing Atlanta, which is helpful for long-distance train travelers from the Northeast going to Florida, but might not have much local demand.
  • The Gulf Coast Corridor, or New Orleans Hub connects Houston to Mobile and Atlanta. This is an official FRA corridor, but seems on a slower track than many of the others, not receiving funding in the most recent round.
  • The long term program includes a line from Phoenix to Tucson (a Phoenix Hub), and from Denver to El Paso (a Denver Hub), but these are both isolated corridors, indicated on the long-term vision, without any likelihood for construction in the short-term. Describing these as hubs stretches the meaning of the term, but those are the primary cities on the respective networks, and are the only cities on their networks with significant feeder public transit. While these local spokes do not show on the national high-speed rail network, they still exist, and support the use of the term for these locales.

Several cities tie together multiple hub networks, these include New Orleans (connecting the Dallas and Atlanta networks as the hub of the Gulf Coast Corridor), Raleigh (connecting the Atlanta and New York networks), Louisville (connecting the Chicago and Atlanta networks), and Kansas City (connecting the Dallas and Chicago networks). Those with an eye to drawing networks would easily conceive of links (not yet on the books) connecting Memphis, Nashville, and Knoxville in Tennessee, or Pittsburgh and Cleveland or Columbus. The unofficial US High-speed rail Association has the most comprehensive network plans, including staging, which includes many of these and other links, see the bottom left on .

These hub networks in the Federal High-Speed Intercity Passenger Rail Program includes the top 47 metropolitan areas of the United States (and many smaller ones), the largest city not on the network is Salt Lake City, Utah, at 48, with just over 1 million people in the metro area 5. The Hubs themselves are metro areas ranked 1, 2, 3, 4, 9, 15, and 27.6 The US High-speed rail Association network includes even more cities.

The political genius of the proposed intercity passenger proposal is that it includes lines in all but 8 of the 50 states7. This is a practice learned in transportation from previous national packages, the Interstate Highway System (with miles in all 50 states (special routes in Alaska and Hawaii), as shown in the bottom right of ) and Amtrak (nearly so), helping ensure strong support in Congress.


(1): Federal Railroad Administration High-Speed Rail Corridor Designations 8.;

(2) US DOT High-Speed Intercity Passenger Rail Program as funded by the American Reinvestment and Recovery Act of 2009 9;

(3) US DOT High-Speed Intercity Passenger Rail Vision 10;

(4) US High-speed rail Association Staging Plan 11

(5) US Public Interest Research Group 12;

(6) Interstate Highway System 13.

Regional HSR Maps

(1) Florida (Orlando Hub)14;

(2) California (Los Angeles Hub)15;

(3) Midwest (Chicago Hub) 16;

(4) Northeast (New York Hub) 17.

The train that never stops at a station (You Tube)

No failures allowed.

Some thoughts on high-speed rail – part 1: Introduction




High-speed rail lines have been built and proposed in numerous countries throughout the world. The advantages of such lines are a higher quality of service than competing modes (air, bus, auto, conventional rail), potentially faster point-to-point times depending on specific locations, faster loading and unloading times, higher safety than some modes, and lower labor costs. The disadvantage primarily lies in higher fixed costs, potentially higher energy costs than some competing modes, and higher noise externalities. Whether the net benefits outweigh the net costs is an empirical question that awaits determination based on location specific factors, project costs, local demand, and network effects (depending on what else in the network exists). The optimal network design problem is hard (in the mathematical sense of hard, meaning optimal solutions are hard to find because of the combinatorics of the possible different network configurations), so heuristics and human judgment are used to design networks.


The network architecture of high-speed rail lines has tended to be in a hub-and-spoke pattern, connecting a hub city (e.g. Paris, Madrid, Tokyo) to secondary cities in tree-like architecture, with occasional crossing links, typically at both lower speed, lower frequency, and lower cost of construction. As these systems were designed nationally, and the largest city is often the capital (as in Paris, Madrid, and Tokyo), which is also (roughly) centrally located, it is no surprise that the hub was based where it was. Germany has fewer very high speed links (faster than 300 km/h), and a flatter (less-hubbed) network, perhaps reflecting its strong federalism, relative decentralization into a multi-polar urban structure and late formation into a nation-state. Italy has centered its hub in Milan, the largest metropolitan area in the country. Maps of International HSR systems are shown in .


The reason for the hub-and-spoke architecture is to achieve economies of density in track usage and network effects at the hub city which enable frequent service to multiple destinations. Multiple paths between origins and destinations would diffuse the network effects and result in less frequent service, and therefore reduce demand. The hub-and-spoke architecture, while benefitting the network as a whole when demand is insufficient to enable frequent point-to-point service, clearly benefit the hub cities the most, as they gain from all the incoming flows which create additional demand, and thus greater service. In air transportation, airlines often use hub-and-spoke networks, and if they have a large market share at a hub airport, will use that advantage to charge a premium for travel, thereby capturing some, if not all, of the benefits of being located in a hub airport city.

Unproductive investments

SFGate: Bay Area transit up for unprecedented overhaul

In the last decade we almost doubled the amount of money we put toward transit, while increasing service only 16 percent and ridership only 7 percent. Which calls for more strong words from the report:

“That is a terrible return on our region’s transit investment, and it should cause us to think long and hard before committing future funds to such a low-yield strategy.”

Read more:

In New York: Transport, Land Use, and Value

I will be in New York on Friday April 30 giving a talk on Transport, Land Use, and Value hosted by the local University Transportation Center at City University of New York.

Given transportation creates land value, and recognizing the problem of underfunding transport infrastructure, new funding sources can be used to increase transport investment, create additional land value, and improve social welfare. This presentation considers co-evolutionary process between the development of land and transport networks. Using data from the rail and Underground in London and the streetcar system in the Twin Cities, the empirical relationship is established statistically under several different contexts, and hypotheses about the positive feedback nature of the interaction are tested.

Time & Location
Baruch College Conference Center
151 E. 25th Street, 7th Floor, New York, NY
9:30 am – Breakfast & registration
10:00 am – Seminar

Small Railroads in Trouble

Oh Dear, The Fat Controller is losing money.
From the Telegraph (such a quaint name for a newspaper) HIT loses $500m in pre-school TV market

“The owner of Thomas the Tank Engine, HIT Entertainment, has been forced to write $500m (£325m) off its value, wiping out the entire investment of private equity owners Apax Partners.”

… blaming a competitive pre-school market. I know our household purchases of Thomasana is down this year, mainly because there is nothing left to buy.
Will Warren Buffett be next?

A puzzling palimpsest of perceived precepts

The lanes on Riverside Avenue in Minneapolis are now a puzzling palimpsest of perceived precepts. Since the snowplows scraped up some of the most recent markings and whatever obscured the previous iteration, Riverside is now not a self-explaining road, instead it is a self-confusing road. Is there or is there not a turn-lane?
Photo taken at intersection of Riverside and Butler Place.

Returning streetcars to the Twin Cities

In Twin Cities Daily Planet: Returning streetcars to the Twin Cities.
A bill before the Minnesota legislature would use value capture techniques (Tax Increment Financing in a Transit Improvement District) to pay for the capital and operating costs of transit, including streetcars.
Whether streetcar benefits outweigh costs for Minneapolis or not, value capture to pay for transportation services is progress (compared with other general fund sources).

Road pricing when? Old fleet, new fleet.

Ryan Avent at The Bellows argues in his post No Choice

“The bottom line is this: we can no longer afford to not tax important negative externalities. We can no longer afford to not do the stuff we really ought to have been doing in the first place. The options we have are to ratchet up current taxes with bad incentive effects and diminishing returns, or to cut spending on important priorities, or both. But cutting back on education spending and infrastructure investment while increasing taxes on income will squeeze growth, making the task of closing these financial holes harder.

Or we can bite the bullet, suck it up, and start charging an appropriate amount for valuable public infrastructure. We can stop giving away space on roads and parking spots for free, costing everyone a lot of wasted time. We can stop letting companies foul the air and slow-cook the earth with no negative impact to their bottom line. And then we just might have enough dough to keep critical infrastructure running. We might even be able to invest in a new and better infrastructure capacity.”

Similarly, the blog Grush Hour argues No more RUC [Road User Charge] trial please, suggesting we know enough and the time for action is now.

While any claims I have to working in transportation economics would be revoked if I did not support road pricing, there are several issues that keep getting ignored in the progressive blogosphere when endorsing quick implementation of road pricing.

  1. Gas taxes send the right signal about general use, and encourage conversion from gasoline to electric powered vehicles if set appropriately, though does not send a useful signal for time and place. See: A Dozen Reasons for Raising Gas Taxes
  2. Gas taxes are administratively efficient, road pricing loses on the order of 20-30% of revenue to administration and collection costs. See: Too Expensive to Meter
  3. Imposing a new mileage based user fee/road user charge/vehicle mileage tax on existing vehicles is going to be unpopular, probably less popular than simply raising gas taxes. See: Road pricing killed off by Transport Secretary

Everyone now recognizes gas taxes will cease to be effective as user charge as hybrid and EV adoption rises (unless the gas tax rises with MPG, and even then 100% EVs will get off the hook). If we rely only on gas taxes, we eventually will have to tax 100% of the cost of roads on the last gasoline powered vehicle. The system will break down long before then. See Beyond the Gas Tax


This suggests an obvious transition point. Use gas taxes to collect revenue from the “old fleet” powered by gasoline or diesel, use a Vehicle Mileage Tax to collect revenue from the “new fleet” powered at least in part by electricity. It can be easily communicated that the new fleet does not consume gasoline (or as much) and this is about fairness. The relative gas and electricity charges can still be skewed to adjust for environmental externalities associated with gasoline, but other than that, should be equalized to reflect costs imposed and benefits received. A standing, independent “Highway User Fee Commission” can set federal rates to ensure full funding of the Highway Trust Fund (and secondarily manage traffic by time and place). States could piggy-back on the apparatus.

Each new electric or quasi-electric car can have an on-board device to compute tolls specific to the vehicle (based on MPG, and therefore discounting for the gas tax already paid) and for time (hour of day, day of week) and general place (in the city, on the freeway, vs. in the country, on local roads, etc.). Since this would apply only to new cars, no older EVs would be harmed (it is a small price to pay for political harmony).

Trucks are another story, since the fleet is smaller and more centrally (though not centrally) managed. They can be converted sooner.