Sunday, July 24, 2011

Delta Airlines trimming service to small cities

Pinnacle Airlines, which operates regional flights as Delta Connect, is retiring a fleet of Saab 340 turboprops inherited from their recent acquisition of Mesaba Airlines. In the process, they've determined that service to 24 smaller airports—including Pierre, the capital of South Dakota—would be unsustainable for them, and have given the U.S. Department of Transportation a 90-day notice that they plan to discontinue or alter service. 16 of the destinations have flights subsidized under the Essential Air Service program, while 8 currently operate without subsidy.

View Delta rural air service cuts, 2011 in a larger map

The table below combines the load factors from Delta's website with subsidy values from Wikipedia's article on the EAS program (some of the subsidy values are out-of-date). The annual passenger estimates are my own, so they could be off by quite a bit (10% or so). I assumed 13 weekly round-trips for EAS airports (6 days with two round trips each and one day with just a single round trip).  Here's what I used for the non-subsidized airports:
Pierre: 2x Sun-Fri, 1x Sat
Sioux City: 3x Mon-Sat, 2x Sun
Brainerd: 3x Sun-Fri, 2x Sat
Aberdeen: 2x Sun-Fri, 1x Sat
Pellston: 3x all days
Bemidji: 3x Sun-Fri, 2x Sat
Waterloo: 3x Mon-Sat, 2x Sun
Butte: 2x all days
AirportCodeEAS subsidyLoad factorEst. pass.BusRail
Thief River Falls, MNTVF$1,230,32212.0% 5,532Nofreight
Greenville, MSGLH$1,606,66227.6%12,725Yesfreight
Devils Lake, NDDVL$1,459,49330.3%13,970NoAmtrak
Watertown, SDATY$1,338,32135.0%16,136Yesfreight
Muscle Shoals, ALMSL$2,553,28335.7%16,459Nofreight
Fort Dodge, IAFOD$1,112,60739.1%18,027Yesfreight
Hibbing, MNHIB$2,938,87839.2%18,073Nofreight
Alpena, MIAPN$1,532,66039.5%18,211Yesfreight
Tupelo, MSTUP$921,87841.0%18,903Yesfreight
Jamestown, NDJMS$1,639,25442.1%19,410Yesfreight
Mason City, IAMCW  $1,112,60745.9%21,162Yesfreight
Pierre, SDPIRNot subsidized  47.4%21,853Yesfreight
Iron Mountain, MIIMT$2,090,53448.7%22,453Yesfreight
Sioux City, IASUXNot subsidized51.4%34,540Yesfreight
International Falls, MN  INL  $1,309,88652.5%24,205Nofreight
Brainerd, MNBRDNot subsidized52.6%37,306Yesfreight
Hattiesburg, MSPIB$1,398,79853.7%24,758YesAmtrak
Escanaba, MIESC$2,090,53455.2%25,449Yesfreight
Aberdeen, SDABRNot subsidized55.6%25,634Yesfreight
Pellston MIPLNNot subsidized58.5%43,559Nonone
Bemidji, MNBJINot subsidized59.3%42,058Yesfreight
Sault Ste. Marie, MICIU$237,82560.0%27,662Yesfreight
Waterloo, IAALONot subsidized61.4%43,547Yesfreight
Butte, MTBTMNot subsidized65.3%32,415Yesfreight
EAS subtotal$24,573,542303,135
Unsubsidized subtotal280,912
Actual passenger counts should be available from the federal government somewhere. The last time I looked into it, getting any useful information appeared to require setting up a huge database.

All of Mesaba/Pinnacle's current EAS-subsidized flights are on the chopping block, and Minnesota is seeing some of the biggest cuts—right across the territory where Mesaba first started flying back in the 1940s. Of eight current commercial airports in the state, five will likely see service reduced or eliminated: Thief River Falls, Hibbing, International Falls, Brainerd, and Bemidji. The first three are currently EAS-subsidized, while the last two don't require subsidy at the moment.

I calculate that the government is contributing about $81 per passenger for the subsidized routes, ranging from less than $9 for Sault Ste. Marie to $222 for travelers to and from Thief River Falls, which has become a bit of a whipping-boy over the last week for being at the top of the list and having the lowest load factor.  Despite Thief River Falls' small size, the city is home to some notable businesses, including snowmobile, ATV, and watercraft manufacturer Artic Cat, and DigiKey, a major distributor of electronic components. It's still debatable whether they deserve airline service, especially since flight time to MSP is slowed by a bounce though Hibbing.  That extra hop makes travel time almost exactly the same as driving to Grand Forks and taking a flight from there.

The other three commercial airports in Minnesota are Duluth, Rochester, and MSP, and they are backed up by airports just across the border in Grand Forks, Fargo, Sioux Falls, and La Crosse. Some passengers already choose to use these airports because of the lower fares that can be found there. In my browsing around, the Pinnacle/Mesaba flights to their central hubs run between $250 and $500, but going to a larger airport can reduce the cost by $70 or more. Annoyingly, these routes tend to exhibit the "hidden city" fare problem quite badly. As an example, I looked at Orbitz for flights late next month from Thief River Falls: Flying one-way to MSP runs $317 (including taxes, fees, and a $25 baggage fee), while a trip ticketed from Thief River Falls through to Chicago would only cost $168.

Delta may continue service to nine of the 24 cities if they receive higher subsidies and/or the money shifts to currently unsubsidized routes. The turboprop fleet is being replaced by 50-seat regional jets, which apparently use more fuel and cost more to operate, so they would need to carry more passengers per plane. Cities that currently see three round-trips per day would most likely see that reduced to two (apparently the prescribed number under the EAS program). That would appear to affect Bemidji, Brainerd, Pellston, Sioux City, and Waterloo.

The EAS program has been a political target for several years now, and could be phased out over the next few years. Representatives in the U.S. House had voted back in May to end the program entirely by 2013 except for Alaska, where the subsidy per town is quite low. The EAS program isn't all that big overall. The subsidy to the lower 48 states is only $170 million per year, which hardly amounts to a rounding error of a rounding error in the federal budget. Alaska runs about $12 million across 45 communities, while the lower 48 states have about 107 EAS-supported routes.

Nonetheless, Delta's actions to cut service to their 24 communities could be seen as a hedge that EAS funding will eventually go away. Why go through the bother of buying replacement turboprops or extra regional jets if they can't be used on their intended routes?

Delta has been in talks with Great Lakes Airlines to see if they'd like to take over some routes.  They've expressed interest in Brainerd, International Falls, Fort Dodge, Mason City, and Iron Mountain, and they already have service between Pierre and Denver.  GLA operates 19-seat Beechcraft 1900 planes, so they could theoretically operate at better margins. 

However, Great Lakes Airlines is also currently in the news because they operate the EAS program's most heavily-subsidized route—a single daily round-trip to Ely, Nevada with a subsidy of $3,720 per head. That's enough to buy every passenger a half-decent used car and probably some fuel and insurance along with that. A political fight over that route and two others that exceed a proposed $1,000 cap is partly why the Federal Aviation Administration went into partial shutdown on Friday night.

AirportCodeEAS subsidySubsidy/passEst. passSeats/Est.
load factor
Ely, NVELY$1,752,067$3,72047119 / 3.4%Nomuseum
Alamogordo, NM ALM$1,169,337$1,5637489 / 17.6%Yesfreight
Glendive, MTGDV$1,056,152$1,35877819 / 2.8%Yesfreight

Ely, NVGreat Lakes Airlines$911x dailyBeechcraft 1900
Alamogordo, NM New Mexico Airlines$752x dailyCessna 209B
Glendive, MTIsland Air$752x dailyBeechcraft 1900

Ely is in the middle of nowhere, so I don't know what the best travel option is for their population.  Clearly a 19-seat plane is overkill for a service averaging one or two passengers per day, so it would obviously be a good candidate to shrink to a small Cessna or something.  They do have a heritage railway—a legacy of the copper mines that helped grow the population in the first place—which appears to have continuous track up to northern Nevada where it meets the California Zephyr route, but any attempt to run trains would certainly be unsustainable (the distance appears to be more than 100 miles). I haven't been able to find any evidence of regular bus service, though that might just mean it's not on the Internet.  Similarly, there might be charter flights available at the airport, which was seeing more than 10,000 yearly operations back in 2007 according to Wikipedia.

The other cities are in much better shape even if they lose the airline traffic. Alamogordo has bus service and is about 90 miles from El Paso, Texas, though that's along U.S. 54, which Google Maps says has restricted access due to military presence along the way. The other way to El Paso is via Las Cruces, which reminds me of an old Wings episode. (Las Cruces lacks commercial service, however.) Glendive has bus service through Rimrock Stages, and  was included as a potential stop along the North Coast Hiawatha route studied by Amtrak in 2009.

I was amazed by the fares when I looked them up on Orbitz. These ones include the taxes and fees, but I doubt they charge extra for checked bags on flights with one or two people.  In contrast to the huge prices I was accustomed to seeing from Delta/Mesaba/Pinnacle flights, these are very low fares.  While the Delta Connect flights would seem to average a farebox recovery ratio well above 50%, these flights are extremely poor, only in the range of 2 to 6%.

Like Alamogordo and Glendive, most of the 24 cities being targeted for cuts by Delta have bus service through Greyhound or another company. Many have daily service, though some routes often only operate three or four days per week. Minnesota again gets hit the worst on that front, since Thief River Falls, Hibbing, and International Falls all lack regular bus service (Hibbing does have a once-weekly jaunt to Duluth and back via Arrowhead Transit, but I think that's too minimal to really be useful). Devils Lake has Amtrak (at least until the lake grows again) and doesn't have regular bus service either, although there is an organization that operates one past town to Churchs Ferry 20 miles away.

Outside of Minnesota, perhaps the most interesting cut is to Pierre, the capital of South Dakota. The current unsubsidized fare to MSP racks up to $506 if a baggage charge is included. The city does have bus service, but it's still pretty far from any other significant airport.  Google Maps estimates the nearest other commercial airport is in Aberdeen, which is also facing cuts, followed closely by Rapid City. In both cases, drive times would be up to about 3 hours.  I'd hate to make the trip by bus or attempt it in the middle of winter when wind-whipped snow drifts across highways and makes them impassable (the snow also becomes a challenge for trains and planes, of course). I guess they'd better hope that Great Lakes Airlines retains their service from the city (which is significantly cheaper anyway).

Only one town out of the whole two dozen, Pellston, Michigan, lacks both bus service and any rails to run some sort of train service, but it appears to have enough passenger traffic that another airline may be interested in the route.

The U.S. House plan for the $1,000 limit per passenger was actually a compromise position—pushed forth instead of cutting EAS entirely—and it generally seems reasonable to me.  I am amazed that the Tea Party contingent was willing to compromise at all, however.  While they hate trains, they're ambivalent about planes, and clearly love cars.  A 90-mile restriction was proposed along with the dollar value, preventing subsidies from being given to cities within a 90-mile radius of a major hub.  It would apparently affect 10 airports, though I haven't been able to find a list.  Depending on the definition, the rule could affect Thief River Falls (near Grand Forks) and Hibbing (near Duluth) up here in Minnesota, and definitely Greenville and Tupelo in Mississippi (where flights go to nearby Memphis).

The $170 million saved by eliminating EAS is still dwarfed by a much larger chunk of money is dedicated through the Airport Improvement Program, which has an annual budget over $1 billion. AIP funding for the next two years apparently amounts to $2.4 billion, while the previous two years split $3.5 billion. I could sure see a lot of trains being built for that kind of money, since it's the equivalent of one to two Northern Lights Express trains per year, 100 to 200 miles of double-tracking at 90 to 110 mph (and that's with our overpriced American signaling, construction, and rolling stock—$1 billion could also cover the upgrades needed for the 2,000-mile North Coast Hiawatha to return to service).

Overall, the Federal Aviation Administration had a budget of $16 billion in 2010, though I'm not quite sure where they get all of it.  They collect $200 million a week in airfare taxes (though apparently not right now due to the shutdown), but that would only add up to $10 or $11 billion per year.

Two cities out of the 24 facing Delta cuts are also served by Amtrak and give us an opportunity to see how ridership compares. The Devils Lake Amtrak station, served by the Chicago–Seattle/Portland Empire Builder, recorded 6,148 passengers (boardings and alightings) in FY2010, while the Hattiesburg station, served by the New York–New Orleans Crescent, had 11,868. Roughly 45% and 48% of the number of airline passengers through the cities' respective airports, and that's with often-delayed service and virtually zero advertising on Amtrak's part.

Both stations see one long-distance train each direction each day, which I generally consider to be half the level of service that EAS airports see.  You can make two round-trips in one day from an EAS airport (albeit sometimes in two opposite directions), but it's only possible to make one round-trip per day on most Amtrak routes.

Here's some more data on those two routes plus a smattering of others that show the range of Amtrak's financial picture.  This is mostly a mix of the best and the worst, though I've also put together a spreadsheet with more calculations here.

TrainRevenueSubsidyPass.Pass.-miAvg. trip
Crescent $30,600,000 $44,100,000298,688164,552,238551 mi
Empire Builder $62,400,000$61,600,000533,493385,000,000722 mi
Northeast Regional$469,600,000$48,800,0007,148,9981,084,444,444152 mi
 Washington–Lynchburg$7,800,000($2,100,000)126,07229,166,667231 mi
Acela Express$449,900,000($100,600,000)3,218,718609,696,970189 mi
Hoosier State$800,000$5,000,00033,6005,208,333155 mi
Hiawatha Service$20,600,000$13,800,000783,06062,727,27380 mi
Texas Eagle$24,400,000$29,400,000287,164166,101,695578 mi
Sunset Limited$11,100,000$39,300,00091,68478,131,213852 mi
California Zephyr$48,300,000$56,200,000377,876302,150,538800 mi
Auto Train$61,700,000$21,600,000244,252211,764,706867 mi

TrainPer passenger tripPer passenger mileOp. ratio
Crescent $102 $148$250$0.186$0.268$0.45440.9%
Empire Builder $117$115$232$0.162$0.160$0.32250.3%
Northeast Regional$66$7$73$0.433$0.045$0.47890.6%
Acela Express$140($31)$109$0.738($0.165)$0.573128.8%
Hoosier State$23$149$173$0.154$0.960$1.11413.8%
Hiawatha Service$26$18$44$0.328$0.220$0.54859.9%
Texas Eagle$85$102$187$0.147$0.177$0.32445.4%
Sunset Limited$121$429$550$0.142$0.503$0.64522.0%
California Zephyr$127$149$277$0.160$0.186$0.34646.2%
Auto Train$253$88$341$0.291$0.102$0.39374.1%

Again, these tables are full of numbers from my own math, so they may be a bit off.  Because of the way I calculated things, error increases as the absolute value of subsidy per passenger mile shrinks toward zero.  "Subsidy" in these tables comes from the "Fully-allocated loss" from Amtrak's October 2010 monthly report

Pretty amazing that our friendly neighborhood Empire Builder generates the third-highest level of passenger-miles in the entire Amtrak system.  Not bad for a once-daily train (though there are usually four trains out plying the rails at any given time due to the distance of the route).

Amtrak's total cost to carry a rider is typically a lot lower than what it costs the airlines, both on subsidized and non-subsidized flights. While airlines were charging $337 per passenger in 2010, the total cost per Amtrak passenger was about $133. Amtrak is at a partial disadvantage due to slower speeds, but a person valuing at their time at $25 per hour would still break even by taking Amtrak at full price even if the train took 8 hours longer (which works out to a minimum travel distance of 400 miles at 50 mph).

The problem is that Amtrak is only charging an average of $64 per passenger right now.  Analysts such as those at the United Rail Passenger Alliance believe that Amtrak could easily charge more, particularly on long-distance routes where fares are currently a lot lower than in the Northeast Corridor. The company definitely should start cranking up prices and find out what the market can bear, though they'd better also be sure to advertise their presence as well.

Some of the fare increases could be entirely invisible, if only the company had the equipment to allow it.  Long-distance trains typically have their sleeping compartments filled months in advance, and those seats are actually charged at or above true market value. I doubt the Tea Party will last much longer, but if it does, Amtrak's annual GAAP loss of $1.3 billion will certainly become a target, no matter what their subsidy per passenger is.

Amtrak's ticketing system already does a modest level of yield management, occasionally increasing fares if trains get filled up.  It's relatively unusual to see fares shifting around, however.  The principle of fare elasticity means that some passengers will likely choose not to travel as the standard fare goes up.  However, I bet the company could get around that problem by using Megabus-style fares, perhaps having the first passenger get a ticket at $0.01 per mile, and subsequent passengers paying more as more seats get filled.

The benefits of  having Amtrak run at or above break-even are too significant to ignore, so the company really ought to bring their ticket prices more in line with actual costs.  EAS subsidies have helped bring about the shutdown of the FAA, and historic rail subsidies have continued to erode the prospects of high-speed rail prospects.  I strongly feel Amtrak could raise prices and winnow down most of their deficit, allowing for better prospects of system expansion and perhaps even reaching some of these EAS cities that will likely see their airlines depart for the last time in short order.

Thanks to Alex for mentioning the airline cuts to me in the first place.


  1. Nice analysis. It's amazing to see towns like Pellston where the airport is larger than the town.

    The URPA makes a pretty good argument that the market could support an increase in fares for long-distance Amtrak routes, but I'd prefer to see it as part of a grand bargain for more capital funding. It's true that people tend to choose long-distance trains for the unique experience, but they will likely only choose it until the first time they're stopped in an Indiana cornfield waiting for passing freights for 6 hours.

  2. Yeah, the general unreliability and low frequency of service is a problem, and is probably the best explanation for why the railroad decides to price long-distance trains at a much lower level than their NEC services. The delta is still pretty huge, though. Frequency and reliability are definitely worth more to me, but I'm not sure if it's a 3x difference.