Fundamentals driving trends in Intercity Transportation in the US

Despite major proposals for investment in high-speed rail, intercity passenger rail in the US remains and will remain a small mode unless there is some sustained exogenous shock to the system such as higher fuel prices, very stringent environmental regulations, an unforeseen war, or act of terrorism that constrains air transportation. As a mode, rail was in decline in the US from a peak in about 1920 through the 1990s. Presently, Amtrak serves 650 million passenger miles in a peak month (a number that has risen considerably since its nadir), US airlines serve about 80 billion revenue passenger miles in peak months).

US Railroads Total KM of Track
US Railroads Total KM of Track

In contrast with the US, internationally, rail passenger service is growing much faster and serves a much larger share of the market. In the UK for instance, there are now 2.8 billion passenger miles per average month (excluding urban transit), well more than four times the US number for a country with one-fifth the population (in other words, rail usage is more than 20 times as high in the UK as the US), a number that has steadily risen, and can expect to rise more with major investments in HS2 and Crossrail.

Lifecycle theory traces out the deployment path of technologies, from birth, through growth, to maturity and decline. These S-shaped curves have successfully described the deployment of many technologies and will be applied to a variety of transportation systems to identify their prospects. The difficulty remains for predicting modes that are still growing, where they will reach market saturation. E.g. how many flights will people take per year? This requires examination of fundamental factors.

There are a number of social and technological changes that may affect outcomes. The most significant for transportation might be automation, in particular self-driving vehicles (what Elon Musk, head of Tesla Motors, recently referred to as “auto-pilot”). (Related automation trends include pilot-less planes (drones) and automated trains).

Self-driving vehicles hold the promise of radically altering urban transportation. Their effects on intercity transportation are less clear. On the one-hand they will extend people’s willingness to travel by auto, as they lower the cost to the driver of travel (in terms of their need to exert energy driving and attending to the road), and enable them to engage in other in-vehicle activities. In that regard, they might change the boundary between the “drive or rail” and “drive or fly” decision (e.g. moving the threshold from 300 miles to 400 miles). Thus, they are more likely to affect rail than flying.

On the other hand, self-driving vehicles will likely decrease auto-ownership and increase various types of on-demand car rental, which I have called “cloud commuting”, such as car-sharing (Zipcar, Car2Go, etc.). This is because one of the major difficulties with car rental, especially in less dense areas, having to travel to get the rental car, will be obviated. People with fewer cars on hand are more likely to use shared transportation modes (transit, intercity rail, airplanes), since they will be paying more per trip (they will have to pay to rent the car, while if they owned, they would not attribute ownership costs to a particular trip).

Trade-offs between modes
Trade-offs between modes

Further, so long as other modes remain faster and less expensive over some distance, longer trips will remain in the domain of train and especially airplanes. In the second figure, user-owned driverless cars will likely shift the location of D1, moving it to the right, while user-rented driverless cars change the fixed cost of making a trip by automobile, moving the intercept of the green line with the Y-axis upward (compared to self-owned cars).

We can use this model to examine other types of shifts as well. For instance, more widespread rail networks will aim to push against this trend, moving D1 to the left, and D2 to the right.

How will these trends play out with current and upcoming technologies, considering physical constraints (such as time availability) and economic constraints (such as incomes)? My guess is that car ownership remains the dominant means of transportation for most Americans, and so the range people will travel by car on the net will increase. But the future is complicated, and other factors may intervene.

NLX redux

I get into a Letter to the Editors battle over in Pine City. I am responding to this about everyone’s favorite intercity rail project, the Northern Lights Express. Keep in mind that the titles of Letters to the Editor are written by the editors, not the author.

Metcalf off the rails on NLX project

Posted: Thursday, May 2, 2013 10:04 am

Professor David Levinson | 0 comments

I am surprised at the Ad Hominem nature of the attack by Metcalf, whom I have never met, but desperate people say desperate things. I won’t respond in turn. I will say though I am trained as an engineer, I have done a significant amount of work in the related field of transportation economics and travel demand modeling. I worked as a demand modeler for 5 years and wrote my doctoral dissertation on the choice of financing mechanisms for highways, which I published as Financing Transportation Networks. I serve on the Editorial Board of the Journal of Transport Economics and Policy. I have over 100 peer-reviewed papers (see, many of them on the topic of transportation economics, and hosted the International Transport Economics Conference in 2009, held at the University of Minnesota. I think I am qualified to write a letter to the editor.

So to begin, what are Metcalf and TEMS doing in this letter if not advocating for the project?

The letter says: “However, the Northeast corridor does not need an operating subsidy as it goes faster than 110 mph. It makes a good operating profit, just as the NLX will if speeds are kept up to 110 mph.” The key word is “operating”, meaning it does not include “capital.” This is a classic shell game. Operating revenue pays for the drivers and fuel, but not the longer-lived trainsets and tracks. Something else must pay for those. Also, the NLX is not the Northeast Corridor and nothing magically happens when speeds reach 110 mph; you get a few more riders in exchange for burning more fuel and paying more upfront for straighter and flatter tracks.

Metcalf writes: “The issue is can the service be franchised, by covering its operating costs and by covering operating costs only. If the train runs at 110 mph the answer is yes, it will cover its operating costs. The state and local communities will only be on the hook for annual operating subsidy not capital costs.”

Yes, state and local communities will be on the hook for the annual operating subsidy. This subsidy is what enables the franchised service to cover its costs. It gets revenue from passengers (farebox recovery) and from subsidy. The capital costs are of course generally paid “up front” by selling bonds. Those bonds will have to be paid back by someone or risk default. Often these bonds are backed by the downstream revenue source of the project they are supporting (they should be, otherwise the public is on the hook for paying them back), if the bond market has confidence that there will be sufficient revenue to pay back the bonds, otherwise it is backed by general revenue. Sure, having someone else (e.g. your Uncle Sam) pick up 80% of the check for the capital costs is better for you than picking that up yourself, that doesn’t mean the cost isn’t paid.

The letter asserts: “The Casino attracts nearly 8 million trips per year with over 3 ½ million visitors or 20,000 trips per day.”

That sure sounds like a big number, so, being an engineer, I looked at the traffic counts (, which counts cars, not people. I-35W has about 20,700 vehicles per day just outside Hinckley in 2010 (total both directions). Near the Casino, there is a count of 9,600 vehicles per day (4,800 in each direction). East of the Casino, traffic counts are 5,100. So from the counts, it looks like the Casino is generating about 4,500 vehicles per day (both directions) or 2,250 vehicle trips into the site (we can annualize this: 2,250*365=821,250, a bit less than 8 million). Now of course, many of those vehicles are workers (the workforce is between 1,000-2,000), so won’t be taking a train. Others are delivery people and so on. So there are maybe 750-1500 vehicles per day of visitors coming in. Some of those vehicles are buses (and have lots of people), some are cars (and carry one to four typically). For what fraction of those people will taking the train be better?

Let’s do a thought experiment: Think about a typical trip-maker, two adults visiting the Casino for some entertainment. They leave their house in the northern Minneapolis suburbs (since there are different casinos serving other suburbs). Will they drive to downtown Minneapolis and pay for parking at the new Interchange Station? (unlikely). Will they drive to a suburban Minneapolis station and park, and pay for two round-trip tickets, and wait for a train, and then have to transfer when they get to Hinckley (since the train is not likely to drop people off at the Casino’s doorstop without a lot of added cost to build a diversion across I-35W), and then the same when going home? (perhaps) Will they drive straight to the Casino? (most likely).

A bigger market are people riding a tour bus, which picks them up at or near their residence. A train can’t beat that, even if it is faster. A train trip also has a lot of slow bits, like getting to the station and waiting for the train, and getting off the train, and going to your actual destination.

So yes, the cursory letter-to-the-editor level gravity model analysis skips the major metropolis of Hinckley Grand Casino and its 1,000 daily incoming vehicle trips. If the train captures five percent of those, its another maybe 100 daily passengers alighting at the station (if private vehicles carried 2 persons each) each way each day, less than 36,500 passengers alighting at Hinckley per year on a line that is purported to carry 10 to 20 times as many.

Next let’s consider the forecasts. We must ask “which forecasts?” Different professional forecasters looking at the line have come up with significantly different forecasts of demand for the line. Notably the State Railway Plan from 2010 (only three years ago) had ridership levels at half what the most recent Environmental Assessment projects. I won’t be getting into why the forecasts might be different (I am sure if you have read this far into the letter, you can surmise, and I suggest you read Flyvbjerg’s Megaprojects and Risk, or his research papers, on the subject for more details.)

The Best forecast from the State Railway Plan for this corridor has just over half the ridership of the more recent Environmental Assessment. The Base case is less optimistic with about one-third the riders.

It has been said that “Gambling is a tax on people who are ignorant of statistics.” Perhaps we should rephrase that … “Investing in transportation infrastructure and hoping for profits is a tax on people who are ignorant of history.”

This is a general problem, not only applied to rail, not just the US, and not only to the NLX corridor. The collective profits from US Airlines from the dawn of the passenger aviation era in the 1930s to the present are negative. The history of railroads is the history of bankruptcy. Minnesota’s storied Northern Pacific went bankrupt in 1875, and again in 1893. James J. Hill’s Great Northern acquired the St. Paul & Pacific Railroad from the bankrupt Northern Pacific and to launch his empire. The beauty of bankruptcy is it wipes out the original investors, but leaves the investment intact. Amtrak, nominally a “for-profit” corporation, was formed in 1970 to relieve freight railroads from their money-losing passenger operations in which they were disinvesting. More recent investments that have failed their initial investors include the Channel Tunnel (reorganized twice), which benefit/cost studies have shown would have been better for Britain if it had not been built (Anguerra (2006)).

There are many other examples, some are described in my book: Garrison, Wm and Levinson, D. (2006) “The Transportation Experience.” Oxford University Press.

See also: Adventures in Forecasting Intercity Rail: NLX edition, Who will pay if NLX fails, Does NLX make sense for Pine County, High-speed rail to Duluth gains steam

Adventures in Forecasting Intercity Rail: NLX edition

Projections around the proposed intercity railway “Northern Lights Express (NLX)” line from Duluth to Twin Cities are presented below. The first two columns of data are from the Statewide Rail plan, funded by MnDOT, prepared by Cambridge Systematics (lead). The last column is from the recent draft Environmental Assessment by USDOT, MnDOT and WisDOT

State Railway Plan Environmental Assessment
Base Best
Scenario Evaluated: High speed, 8 RT High speed, 8 RT
Phase I I Route 9
Capital Cost $878,500,000 $676,600,000 $820,000,000
Operating and Maintenance Cost (Annual) $45,700,000 $35,900,000
Revenue $9,600,000 $12,000,000 $27,660,000
Farebox Recovery 21% 34%
Capital Cost per Mile $5,800,000 $4,500,000
Capital Cost per Rider (2030) $2,042 $1,049 $732.14
Operating Subsidy per Rider $83.82 $36.96
Ridership 2020 938,000
Ridership 2030 430,000 645,000 1,120,000
Ridership 2040 1,302,000
Feb-10 Minnesota Comprehensive Statewide Freight and Passenger Rail Plan
Apr-13 Northern Lights Express High Speed Passenger Rail Project from Minneapolis to Duluth, Minnesota

Governor Maps Road to ‘Interstate’ 41

WBAY news in Green Bay reports the “Interstate” is now a valuable brand. Governor Maps Road to ‘Interstate’ 41:

May 04, 2011 12:35 AM EST
By Kristin Byrne
Big changes are in the works to turn U.S. Highway 41 into an interstate, which Governor Walker says could mean more economic growth in the state.
Governor Walker was in Green Bay Tuesday to make the announcement.
The interstate would run 142 miles, from the Mitchell Interchange on I-94 in Milwaukee to where Highway 41 connects with I-43 in Green Bay.
So how do you turn a highway into an interstate? And what does it mean for drivers?
Transportation officials at the news conference said drivers won’t see much of a difference but the change will ignite economic growth because businesses will want to locate near an interstate.
‘Everybody thinks the interstate is like a brand, and then that brand says access, it says safety, it says speed, it says convenience. And we are essentially, you know, in this corridor, we’re affiliating ourselves with that brand,’ Transportation Secretary Mark Gottlieb said.
‘You’ve got an issue that makes it easier for cars and trucks to drive on. It’s certainly safety factors are built in which makes it a more safe corridor, and long term, not only for drivers, it will be good for businesses because it’s going to be easier to get on and off,’ Governor Walker said.
The governor and the Secretary of the Department of Transportation explained the project could include widening shoulders, making median improvements, and replacing crossroads with on- and off-ramps.
Some preparation work has already been done on Highway 41 with improvements in Brown and Winnebago counties.
The upgrade could cost between $15- to $20 million, mostly federal money the DOT says would come from existing funds.
‘What we do is use it on focusing on the infrastructure that, for really for many of the last several years, has been largely ignored,’ Governor Walker said.
The goal is to have interstate signs up in 2015.

This would be interesting to validate, do otherwise equivalent roads draw different traffic or development simply due to the “Interstate” brand, or is this just hooey (a technical term)?

Trains, Planes, and Automobiles



trainsandplanes-thumb-400x465-73719Paul Krugman (who famously models transportation as an iceberg (and he got his “Nobel” prize in economics for his work in spatial economics and trade theory, showing how aspatial the field is in general), writes about: Trains, Planes, and Automobiles.

There are several problems with this image. First it assumes you are already at the train station waiting to board, as opposed to somewhere randomly in the region. Remember most people do not work or live downtown (even in New York City). Second it ignores the third mode of the title (automobiles). A redrawn figure, which is standard in transportation economics or geography, is shown below. [Similar graphs apply to freight, just change it to Trucks, Trains, and Ships]. The question is whether there is a range between d1 and d2, that is, does rail actually dominate both autos and planes over any region. In terms of travel time it probably does, and looking only at operating cost, it might. In terms of overall cost, including the fixed cost of construction of a new HSR line, it probably does not under current cost structures. The size of this range, if it exists, is, however, empirical, and subject to change with costs and technologies.


A Political Economy of Access: Infrastructure, Networks, Cities, and Institutions by David M. Levinson and David A. King
A Political Economy of Access: Infrastructure, Networks, Cities, and Institutions by David M. Levinson and David A. King


Intercity Bus: Jefferson Lines

Strib has an article on Jefferson Lines, a local Intercity Bus company: Jefferson Lines: Flying low

Traffic on daily roundtrips has risen from 16 passengers a day between Duluth and Minneapolis when Jefferson took over from Greyhound in 2004 to about 90 a day. And that’s before Jefferson added two more express routes last week.

Compare with the proposed so-called High Speed Rail (110mph) Northern Lights Express which will cost at least $615 M (and probably closer to $1B) to build, serving the same market.

The American Bus Association Foundation commissioned a study by DePaul University Prof. Joseph Schwieterman that found that motorcoaches are the most fuel-efficient transportation mode. They provide, on average, 207 passenger miles per gallon, compared with 27 miles per gallon for single-occupant automobiles, 44 miles per gallon for airlines and 92 miles per gallon for commuter railroads, based on typical passenger loads.
What’s more, the airlines need something more than 80 percent of seats full to make money. [Jefferson Lines] needs fewer than half the seats filled on a 55-passenger coach to make a profit.

So the NLX will travel 155 miles in 2 hours (averaging 77 MPH), while buses, which are slightly slower (we could say they should be able to average 60 MPH on this route, which is almost entirely interstate), but more fuel efficient. So the difference is 0.58 hours. Is 17 MPH for 2 hours (i.e. 34 minutes savings for 90 people a day worth a $1B capital investment and a higher environmental cost? 90 people * 34 minutes/person * 365 days/year * 30 years = 33 507 000 person minutes saved. (or $30/minute or $1800/hour). That is a high value of time required to justify this HSR line over its most similar alternative, commuter bus. (I.e. car is not the right comparison, since most people are going point to point, not station to station, and will need transportation on the other end. People going to/from Duluth to the MSP airport to change planes, will probably still want to fly rather than transfer in downtown Minneapolis at significantly extra time)
(This back of the envelope analysis ignores discounting (which makes the value of time higher) and induced demand (which makes the value of time smaller))
OK, so bus is not marketed the way HSR would be, and HSR might attract more passengers. But surely this problem is solvable for significantly less cost than the capital required to construct a new HSR line.

Eisenhower Interstate System in the style of H.C. Beck’s London Underground Diagram

Via Yglesias: On Flickr (may require login) Eisenhower Interstate System in the style of H.C. Beck’s London Underground Diagram

Do Americans Really Hate Flying? Or Really Love Driving?

From FiveThirtyEight: Do Americans Really Hate Flying? Or Really Love Driving?, an amateur analysis of inter-city travel statistics that includes a discussion of gravity models. The comments are worth reading as well.