Wednesday, March 6, 2013

Amtrak Route Performance: Right Numbers, Wrong Analysis

A few days ago, the Brookings Institution released a report entitled, "A New Alignment: Strengthening America's Commitment to Passenger Rail", by Robert Puentes, Adie Tomer, and Joseph Kane. It crunches the numbers on ridership and cost for all of Amtrak's routes.

Many of my friends in the passenger rail support community are deeply disappointed and even angry about this report. "It savages long-distance routes," says Dr. Bill Pollard, President of the Texas Eagle Marketing and Performance Organization (TEMPO), which has very effectively supported its namesake long-distance route for at least 25 years. The National Association of Railroad Passengers (NARP) was somewhat more measured in expressing its disappointment, giving Brookings the thumbs-up on some of its points, and thumbs-down on others.

The report focuses on financial aspects, and certainly most of the criticism in Congress and elsewhere has taken that approach. But the report points out as well that other modes of travel are also subsidized by Federal and state funds, and the financial benefit of transportation - especially passenger rail - needs to be considered much more broadly. And they did not fail to mention that Amtrak's long-distance routes provide service to parts of rural America that have been abandoned by both air and bus lines.

Now, I respect the Brookings Institution and its scholarly approach to issues of national importance. Puentes, Tomer and Kane did a good job of gathering the numbers, and many of their conclusions make a lot of sense. I sincerely doubt they were trying to "savage" the long-distance routes, but chose comparative metrics that don't tell the full story of long vs. shorter routes.

One of their main recommendations is to get the states that want long-distance routes involved in supporting them financially - a recommendation that seems to make sense, as we'll see later in this post.

Not so fast!

Puentes and his colleagues may have looked at a lot of numbers, but not always in the right way. They divided Amtrak routes into two categories: those up to 400 miles, and those longer than 400. Although that makes a lot of sense superficially, it ignores underlying factors which have much more to do with cost and ridership than just the length of the routes. The factors of real importance to determining a route's success are travel time, frequency, and reliability.

Each of these three factors happen to align fairly well with route length, and it's very easy to run correlations between miles and dollars. But it's a big mistake to stop there. Each long-distance route has longer travel time, at most one train a day, and mediocre to poor on-time performance.
Acela trainsets being serviced at Sunnyside Yard, Queens, New York
Here's a good illustration: As we all know, the route that gives Amtrak its best financial return is the Northeast Corridor (NEC). It also provides the shortest downtown-to-downtown travel time compared to auto, bus and air; the greatest frequency of service; and very good on-time performance (85% overall). To be fair, the Brookings report mentions each of these factors; it just doesn't extend them into its conclusions about long-distance routes. But the Northeast Corridor is actually 453 miles from Boston to Washington. An Acela Express leaving at 5 AM arrives in Washington at 11:47, 6 hours 37 minutes after leaving Boston - an average of 68.5 MPH including station-stops. These Acela trains charge premium prices, yet are heavily patronized; there are ten Boston to Washington Acelas each weekday, and they were on time 92% of the time in 2012.

Brookings contrasts these with long-distance routes: "Every single one of the eighteen corridors traveling longer than 400 miles operated at a negative operating balance in 2011, whether traveling just over 400 miles on the Pennsylvanian or clear across the country on the California Zephyr". Now, the Pennsylvanian travels a distance of 444 miles, actually less than the distance from Boston to Washington. But it takes 9¼ hours get from New York to Pittsburgh - an average of 48 MPH - and there is only one daily train each way, though like the Acelas it was on time 92% of the time in 2012.

I'd be more than willing to let Puentes off the hook for treating the NEC as less than 400 miles - it is close - had he not pointed to the Pennsylvanian as losing money for being over 400 miles, when its route is shorter than that of the Acelas. That's pretty sloppy. So let's look at each of the factors I believe are truly important, and illustrate them with cases that show their impact.

Travel time

The distance measure of 400 miles is arbitrary, but it does correlate to a fairly steep drop-off in ridership on many routes. But as I just pointed out, route-miles are not as important as travel time. Rather than base conclusions on the route-length of a train, our conclusions should be based on the travel time, particularly in comparison with other travel options.

This is important - not just an academic distinction. When we use time rather than distance, we come up with entirely different solutions to problems. Not much can be done inexpensively to shorten the rail-distance between two destinations, but several things can be done about the time. For example, shortening or eliminating en-route delays, both at stations and at sidings; using rolling-stock that can negotiate a route without slowing down so much at curves; building relatively short segments of track (say, between one and ten miles) that enable passenger trains to enter and leave cities more directly and with less delay. Another tactic is to add a second station track and platform at certain locations where passenger trains either block one another or impede freight movements on host railways. These are strategic investments that are not as glamorous as increasing the top speed on tracks in open country, but could cut significant time from many routes.

Case in point: West Detroit Junction. This one choke-point is responsible for ten minutes' delay between Detroit and Dearborn, Michigan. The distance is only 9 miles, but the current schedule has to allow 27 minutes between these two stations - an average speed of 20 MPH, unacceptably slow. By adding a pair of switches and a short connecting track (work is about to begin on this), trains are expected to take only 17 minutes, an improvement of 37% without changing speed limits or buying new, faster trains. Six Amtrak Wolverine trains make this run every day on their way between Pontiac Michigan and Chicago, 365 days a year. This seemingly small savings - ten minutes - adds up to 60 minutes each day, or 365 hours each year.
Another case: The Brookings report compares the Hoosier State with the Hiawatha service between Chicago and Milwaukee:
However, not every short-distance corridor benefits from such large metropolitan anchors. The Hoosier State runs between Chicago and Indianapolis, a similarly-sized anchor to Milwaukee on the Hiawatha. At double the distance and only one daily departure the Hoosier State's 2012 ridership (37,249) was just four percent of the Hiawatha's (819,493). Partly as a result, the Hoosier State lost over $100 per rider in 2012.
The Hiawatha service runs on straight, high-speed track. With three intermediate station-stops, it completes its 86 mile run in one hour thirty-seven minutes during evening rush hour, an average speed of 53.2 MPH, which is hard to match on Chicago's congested expressways. Meanwhile, the poor Hoosier State navigates a slow, tortuous route through southern Chicagoland. It arrives at Indianapolis six hours and five minutes later, having completed its 196 mile run at an average 32.2 MPH - a speed easily beaten by any car or bus on largely rural Interstate 65. The Brookings report mentions the extra distance on the Hoosier State, but to understand the difference, one needs to consider the time - nearly four times as much.


Empire Builder in Minot, North Dakota.
No long-distance train in Amtrak's schedule runs more than once each day. This makes service to some parts of the country very inconvenient, coming only in the wee hours of the morning. No wonder ridership is low and expenses high!
Lack of frequency also results in a disproportionate cost per passenger. Stations along routes with infrequent service cost almost as much to maintain as those with lots of trains. It is also more difficult for Amtrak to utilize train equipment fully when trains don't run as frequently - especially when they run only three times weekly.

Case in point: The Texas Eagle runs daily between Chicago and San Antonio Texas (thanks in large part to TEMPO), but the Sunset Limited runs only triweekly between New Orleans and Los Angeles. In San Antonio these trains meet; cars from the Eagle are exchanged with the Sunset to provide a Chicago to Los Angeles through service. In order to make this work, two Superliner coaches are left in San Antonio between Sunset runs. The result is equipment under-utilization, and when monetized against to the Eagle's performance, makes it look more inefficient than it really is.


Long-distance trains as a group are less reliable than short-distance trains for an obvious reason: the further they travel, the more things can go wrong. The problems can be either Amtrak's or the host railroads' - or neither, like the weather. In order to plan for these problems, certain stations have long dwell-times ("schedule-padding"), causing travel time to take a hit.

Case in point: the Blue Water, running daily between Chicago and Port Huron, Michigan. Though it only runs 319 miles, Amtrak and the Michigan Association of Railroad Passengers have recently received many complaints that the train arrives at station after station and sits there waiting for 15 or 20 minutes. Why? Because in previous years the host railroad (CSX Transportation) had been working on tracks and otherwise slowing down the Blue Water, resulting in very unreliable schedules. Now that it's running on time nearly every day, CSX doesn't want to spoil its record of delivering the train to Chicago on time by shortening the schedule.

One long or many shorts?

One other way of analyzing long-distance trains was overlooked by the Brookings report. Yes, you can consider the California Zephyr as a single run from Chicago to Oakland/Emeryville, California. But as NARP and the Midwest High Speed Rail Association have pointed out more than once, you'll understand how they really function if you look at them as a large series of shorter connections. The Zephyr connects the Bay Area with Lake Tahoe, Reno with Salt Lake City, and Denver with Omaha. Not to mention connecting smaller towns, like Reno to Winnemucca, Helper Utah to Grand Junction Colorado, and McCook Nebraska to Ottumwa Iowa. In effect, each long-distance train is a whole series of trains that run less than 400 miles each. True: these smaller routes are not the large metros that generate so much of Amtrak's business. But the small communities are in many respects the "heart" of the United States, and are being abandoned by all the for-profit carriers, both air and highway.

And though these small towns and cities may not contribute much "fare-box" revenue, they are important to Amtrak's political survival - in fact, they are vital. Time after time, anti-Amtrak presidents have submitted zero-dollar Amtrak budgets to Congress. Time after time, senators and congressmen from both parties have saved Amtrak because of the small towns they represent. The TEMPO group is made up almost exclusively of citizens from small towns along the Texas Eagle's route. Why? Because they believe whole-heartedly that this one train is vital to their town's economic well-being.

Now what?

Despite its analytical oversights, the Brookings report makes a very important suggestion: involve the states in financing the long-distance trains that serve them. The report points out that with "skin in the game", states work harder to make sure they get quality rail service. One can hope that before too long, states with only "midnight service" will cooperate to get a second frequency along their line, providing daytime service for their residents. Remember: it doesn't cost twice as much to add a second frequency, because the stations and support facilities are already in place. (The real challenge is convincing the freight railroads to host them.)
The report says, "When states contribute to Amtrak operations, they have a vested interest in service quality". A couple of small examples:
  • Over the last six years, Illinois has allocated significant funds to provide two extra round-trips from Chicago on each of three routes served by Amtrak long-distance trains: to St. Louis (284 miles on the route of the Texas Eagle), to Carbondale (309 miles on the route of the City of New Orleans), and to Quincy (258 miles, mostly on the route of the California Zephyr), each taking about 5½ hours. In response to requests from Illinois riders, the state negotiated a special menu for the cafĂ© on state-funded trains.
  • Wisconsin pays Amtrak to run the seven daily Hiawatha round-trips between Chicago and Milwaukee. According to Chicago Union Station staff, Wisconsin DOT is very sensitive to tardiness in getting Redcaps to the Hiawatha platform, and will insist to Amtrak management that service be provided promptly upon train arrival.
A larger example of the effectiveness of state investment of resources is the Midwest Regional Rail Initiative, formed in 2004. The eight states in this coalition agreed on region-wide primary routes and prioritized improvements to be made. Their cooperation landed them large sums of ARRA funding when that became available (though of course Wisconsin and Ohio suffered political setbacks at that time). The funding resulted - among other things - in a joint order (with California) for new railcars and locomotives, the upgrading of track speed in Illinois to 110 MPH between Chicago and St. Louis (in progress), and the state purchase and future upgrading - with federal assistance - of the line between Kalamazoo and Dearborn in Michigan.

So state involvement is a good thing, as the Brookings report points out. Given the lack of effectiveness currently evident in our national legislature, I have to disagree with NARP's press release, stating that long-distance trains should remain solely a federal resonsibility. We must certainly insist on continued federal support for long-distance service, but the states are probably the best level on which to advocate for better rail service - both for long-distance and corridor trains. Let's go for them both.
Amtrak California: state-supported San Joaquin at Bakersfield
To learn more:
* All links accessed 2013-03-16


  1. this good post


  2. I would like to thank you for the attempts you've obtained created in writing this report. I am hoping the same best operate from you inside the likely also. In fact your inventive creating expertise has impressed me to commence my personal Website Engine weblog now.
    limousine service in Dallas, Texas
    Dallas Limousine service