Note: I accidentally posted an incomplete (and incorrect) version of this post yesterday and subsequently deleted it; it may have shown up on RSS readers, and I apologize for any confusion.
Edward Glaeser “ran the numbers” on high-speed rail in the New York Times this week, hoping to provide some insight into the value of investing in a fast train network in the United States. His analysis, focusing on a hypothetical 240-mile line between Dallas and Houston, concluded that annual capital costs (including back payments on initial construction) would be vastly higher than benefits added from the completion of the project. His piece’s conclusions weren’t particularly compelling.
Ryan Avent responds well in a piece for Streetsblog, arguing that Glaeser’s simplistic analysis doesn’t take into account a number of important variables that may well influence the system’s future use. Significantly, no consideration is made for the land-use implications of rail investment. Is Glaeser ignorant of the fact that rail stations located downtown reduce automobile use by encouraging the development of dense, pedestrian-friendly communities around them? Does Glaeser not realize that high-speed rail, operating at much higher capacities than airplanes and therefore offering cheaper prices (if it’s managed well), could not only take over the air market, but also displace a significant percentage of existing road travel?
It’s worth considering whether Glaeser’s assumptions about ridership — 1.5 million trips a year — make much sense. Here’s a comparison between arguably the most successful high-speed rail line in the world (the French LGV Sud-Est) and the one Glaeser hypothesizes in Texas:
| Center City 1 Pop | Center City 2 Pop | Metro 1 Pop | Metro 2 Pop | |
| Paris-Lyon | 2.1 m | 0.4 m | 11.1 m | 1.6 m |
| Dallas-Houston | 1.3 m | 2.2 m | 6.3 m | 5.7 m |
It’s true: the French cities are more densely populated and their public transportation ridership is higher. But evidence suggests that high-speed rail investment is effective in inducing significantly higher demand on corridors between huge cities such as that between Dallas and Houston. In 1980 (before the Paris-Lyon line opened), 12.2 million people rode the conventional rail line between the cities; by 1989, 22.1 million were riding the TGV between the two. That number has likely almost doubled today, considering that overall TGV network ridership is at about 100 million a year.
Despite cultural differences, can Glaeser really support his claim that only 1.5 million trips will be made between Houston and Dallas given the comparison with the French line shown above? Is the American psyche that incompatible with international experience?
The most serious flaw in his piece, however, was his unwillingness to compare the cost of high-speed rail with that of expanding airports and relieving congestion on roadways. As Robert Cruickshank often puts it on the California High Speed Rail Blog, “The cost of doing nothing is not zero.” In other words, fast trains may be expensive to get running, but in many cases, their implementation costs are far lower than those of building new interstates or air terminals.
No matter the costs of high-speed rail, then, they must be measured against the equivalent costs of investing in road and air travel, both of which are just as expensive and far more troubling from a number of other, non-economic perspectives. Mr. Glaeser’s numbers need to be run again.
11 replies on “On Numbers and High-Speed Rail”
Great post.
It’s probably true that if you study Dallas and Houston as those cities are today, you can argue for relatively low numbers, because these are sprawling cities where a high percentage of HSR trips would need to be completed by car due to the lack of transit networks or land use patters that could support them. There’s something funny about looking at distance and population comparisons of Dallas-Houston alongside Paris-Lyon, as though Houston’s HSR riders could get to a single departure station as easily as those of Paris can.
In your last paragraph, note that airport runways, not terminals, are generally the critical limiting factor. Many urban areas have no practical option for further runway expansions or new airports, even if such a project could cover the massive costs of noise mitigation for the inevitable neighbors.
Yonah – I appreciate this response to Ed Glaeser’s piece. Despite my strong support for HSR in many places in the U.S. East and Midwest, I too struggle with HSR as a first stage investment for the 20th century cities designed around the automobile (Dallas and Houston being two classic examples).
You concede: “It’s true: the French cities are more densely populated and their public transportation ridership is higher.” May I suggest that “higher” is perhaps is an understatement? My impression of Dallas is that its CBD a virtual ghost town after the close of business. Small pockets of vibrancy can be found in neighborhoods radiating for nearly 100 miles in any direction.
Further, I would begin to wonder how over-crowded the Dallas-Houston air and surface corridors actually are. Is this pair comparable to SFO-LAX or LGA-DCA?
Should our Federal investments in these regions that have made the ‘mistake’ of sprawl be better spent FIRST on infrastructure that fosters urbanization – including construction of heavy rail/metros.
Frankly, I have the same concerns about CA-HSR in Southern California. I feel the plan focusing on Union Station in downtown along makes a lot of assumptions about rider willingness to inconvenience themselves. The city is gridlocked. I’m concerned about the lack very efficient connections to lucrative rider markets on the west side (Santa Monica) / South Bay (Long Beach) and San Gabriel Valley (Pasadena). LA is making strides but she’s not there locally yet with regional rail. Speed enhancements, removal of grade separations and westward expansion of Metrolink would be a good place to start.
My ask, there needs to be bold local regional transit strategies in place before Federally funded HSR investments are made in each of these sprawl markets.
Even in the Sunbelt, US cities often have star-shaped commuter rail networks centered at the intercity train station. LA’s overall regional rail network is bad because it doesn’t get you to where you work, but it does get you to Union Station. The same is true for DART. MARTA gets you to the former central train station, which is still there albeit abandoned by Amtrak. Those commuter rail systems are essentially HSR feeders constructed decades before the HSR itself.
In any decision process, it is impossible to make an intelligent decision without examining the relevant alternatives. Unfortunately this is often lost even on economists who never thought to take an introductory course in management science. In the case of regional high-speed rail, alternatives include additional intercity air service, intercity bus, and greater highway capacity. Expanding regional air service does not compare favorably with HSR. Intercity bus service is cost effective but does not scale and cannot achieve the economic benefits of HSR, and additional interstate capacity is very expensive and promotes the same old land use policies, exacerbating urban congestion, carbon emissions, and poor air quality. Go with the solution that promotes redevelopment and economic vitality at the city’s core. High-speed rail.
It does seem worth pointing out that there is no proposal to actually build a direct Dallas-Houston HSR line. The current official Texas corridor doesn’t even include Houston. The likely corridor to be adopted, the so-called Texas T-Bone, has a main line connecting Dallas, Waco, Ft. Hood, Temple, Austin and San Antonio and a shorter line running from Houston through College Station to meet the main line at Temple.
One assumes the reason there is no planned line directly connecting Houston and Dallas is there wouldn’t be sufficient ridership to support it. The point of the T-Bone main line is that it attracts riders from any of the cities it connects to any other. The problem with Houston-Dallas is there aren’t intermediate stops between them. Such a line would have to be justified solely on the potential traffic between the endpoints. It is very rare that you can do that.
Glaeser is literally attacking a strawman.
Both Dallas and Houston are expanding their light-rail networks. Dallas will have four light rail lines by 2013 and Houston will have several as well by that same time. And those systems are connected to the central rail terminal, so there are local transit options available in both cities.
europe, much like the eastern half of the US is also much less concentrated than on the western half of the US. europe and the eastern US have lots of small towns scattered around every few miles in all directions throughout the whole state. go out west and the population is almost entirely concentrated along a few travel corridors (i.e. rail lines). no one lives in Nevada outside the las vegas area and reno/carson city area and the few that do then live in tiny communities along the transcontinential RR line like elko, nv. but 90% of the land area in the state of NV has no one on it. the same with OR, WA, UT, AZ, CO, etc. the population out west is in metropolitan areas or else along a few major rail lines (which populated the west in the first place). this is very obvious when you look at a night space photo of the US and see where the lights are.
http://antwrp.gsfc.nasa.gov/apod/image/0007/usanight_dmsp_big.gif
Ray, you’re putting too much concern that high-speed rail is dependent on European or Japanese level local transit in order to bolster ridership.
If you look at the Amtrak California services right now, all 3 routes are in the top 10 nationally. Added together, the Pacific Surfliner, Capitol Corridor and San Joaquin are half of the Northeast Corridor in daily ridership. This is with regular-speed, 79 mph maximum rail available on a dozen or so runs per day.
Furthermore, if rail ridership is dependent on local urban transit, you would expect the Capitol Corridor in the Bay Area and Sacramento to be higher in ridership than the Los Angeles-San Diego corridor. No Amtrak train connection to San Francisco is immaterial; BART and the Thruway buses from Emeryville solve that problem.
Yet Southern California still produces higher ridership, and has always led in ridership even while the San Diego and L.A. rail systems were still much smaller in scope. The difference, though, is that L.A. has long had highly extensive and frequent bus service. San Diego did not until only about three or so years ago.
This also doesn’t explain the relative success of the San Joaquins. There’s no direct train service to L.A. (can’t cross the Grapevine), it takes the most roundabout route to the Bay Area (a hook from Stockton to Martinez to Oakland), and it runs through predominantly rural counties with post-World War II suburban sprawl defining the urban cities. The best bus service is in Fresno, and that’s 30-60 minutes with service shutting down at dusk. Nationally, though, the San Joaquins are around 6 or 7 on the Top 10 list.
The first stage investment for cities built for the Automobile when you have large enough cities within 300 miles of each other is Emerging HSR – 110mph, upgraded level crossings, some targeted bottleneck eliminating grade separations (both road/rail and rail/rail).
If the same GAO report that Glaeser used for a ~$40m/mile mid-range construction cost estimate of Express HSR is used for Emerging HSR, its ~$8m/mile, ranging from $4m to $11m depending on terrain and other alignment specifics. That puts the annual cost at 5% real discount (quite high for publicly financed infrastructure) at ~$80m-$90m/year.
Even under Glaeser’s assumption of “business as usual except population growth comes to a screeching halt”, and given that Dallas and Houston are far from the first priority in potential rail corridors, that brings the benefit/cost ratio very close to 1.0 … if there is any population growth, which will substantially increase the benefit of new transport capacity, it is easily over 1.0 even in a stovepiped analysis that ignores the substantial third party benefits and costs of any big transport project.
With a normal mix of downtown and outer suburban stations for an HSR line, the rail line is more accessible than the airport for suburbanites as well as offering the main downtown station as an anchor for building up the local urban dedicated transport corridors.
Finally, since well chosen HSR corridors … both Emerging HSR and Express HSR … can generate operating surpluses, once established. That makes them an excellent anchor to leverage more off-peak transportation potential for suburban areas, increasing the number and quality of services that can be supported outside the urban core, and thereby providing the foundation for expanding the political coalition in support of local public transport.
“No matter the costs of high-speed rail, then, they must be measured against the equivalent costs of investing in road and air travel, both of which are just as expensive and far more troubling from a number of other, non-economic perspectives. Mr. Glaeser’s numbers need to be run again.”
No matter what the cost….? I don’t necessarily disagree that HSR might be a good investment, but it’s bogeymen logic that we would use all that money for highways instead. There’s no way that we’re going to see additional highway investment at the scale we’re talking with HSR in California. We can’t even get the last leg of the 710 built. So your analogy here is a little like me saying “oh,it’s a smart plan to eat a box of Twinkies because it’s better than the 14 cheeseburgers I could have eaten.” Um, no–that does not good policy make. There are two separate analytical questions in play: 1) do we need additional mobility between metropolitan regions and 2) if so, then where should we put our money? I’m unconvinced that the answer to #1 is a yes.
California’s well-publicized budget deficit and standoff occurred a mere few months after California voters approved nearly $10 billion in obligations for its high-speed rail project. In the end, California’s budgetary crisis was resolved with no tax increases but steep cuts to social programs and prisons. One round of cuts that took over $6 billion from K-12 public schools, and $1.3 billion from the state’s medical program for impoverished Californians, including a million children. Additional cuts removed another $80 million from child welfare programs. The state went overnight from among the most social support for the impoverished to having among the least.
So the question becomes is our HSR dream coming at the expense of scary-scary new highway investment–as you would suggest–or is it coming at the expense of other social priorities in world where fiscal capacity is not infinite?
Note I wouldn’t be kvetching here if we had some sort of solid plan in place for changing the state’s fiscal capacity in order to pay for HSR, such as a good business plan (unlike the one we have) for HSR, a penny increase in the gas tax, or a modification in Prop 13.
On the New Republic‘s The Avenue blog this week, an article by Jonathan Rothwell of the Brookings Institute, reruns Glaeser’s analysis with different (more accurate, he writes) inputs:
http://www.tnr.com/blog/the-avenue/high-speed-rail-getting-the-assumptions-right