Saturday, November 9, 2013

Amtrak routes by 2013 cost recovery

As I did last year, I now present Amtrak's routes ranked according to cost recovery, color coded between the Northeast Corridor, state supported, and long distance trains. As with last year, while the Acela, Northeast Regional, and long distance trains use total revenue, the revenue for the state supported trains (in green) is solely based on ticket fares. While this does depress their actual performance somewhat by a few percentage points (from food and beverage sales and hauling private varnish), it eliminates the effect of state subsidies.


Revenue 2013
Expenses 2013
Cost recovery 2013
Revenue 2012
Expenses 2012
Cost recovery 2012
Change in cost recovery
Change in revenue
Change in expenses
Acela
$542,700,000
$305,800,000
177.47%
$521,000,000
$312,200,000
166.88%
10.59%
4.17%
-2.05%
Washington-Lynchburg
$11,744,966
$8,700,000
135.00%
$11,411,821
$8,100,000
140.89%
-5.89%
2.92%
7.41%
Northeast Regional
$585,800,000
$452,800,000
129.37%
$552,000,000
$476,400,000
115.87%
13.50%
6.12%
-4.95%
Washington-Newport News
$32,916,626
$31,600,000
104.17%
$34,286,847
$31,700,000
108.16%
-3.99%
-4.00%
-0.32%
Washington-Norfolk
$6,233,871
$6,100,000
102.19%
N/A N/A




Carolinian
$19,841,847
$22,900,000
86.65%
$18,652,552
$20,500,000
90.99%
-4.34%
6.38%
11.71%
Keystone Service
$35,442,502
$47,200,000
75.09%
$32,970,951
$47,600,000
69.27%
5.82%
7.50%
-0.84%
Maple Leaf
$23,796,560
$33,000,000
72.11%
$24,600,726
$28,400,000
86.62%
-14.51%
-3.27%
16.20%
Auto Train
$75,400,000
$107,300,000
70.27%
$74,100,000
$108,500,000
68.29%
1.98%
1.75%
-1.11%
Hiawathas
$16,287,184
$25,400,000
64.12%
$15,963,261
$26,900,000
59.34%
4.78%
2.03%
-5.58%
Pennsylvanian
$10,431,324
$16,300,000
64.00%
$9,281,813
$15,600,000
59.50%
4.50%
12.38%
4.49%
Empire Service
$44,299,328
$71,100,000
62.31%
$43,877,344
$65,900,000
66.58%
-4.28%
0.96%
7.89%
Palmetto
$19,100,000
$33,900,000
56.34%
$18,400,000
$29,600,000
62.16%
-5.82%
3.80%
14.53%
Pacific Surfliner
$62,576,548
$113,800,000
54.99%
$58,595,820
$118,900,000
49.28%
5.71%
6.79%
-4.29%
Empire Builder
$72,900,000
$133,200,000
54.73%
$72,200,000
$129,800,000
55.62%
-0.89%
0.97%
2.62%
Downeaster
$8,211,723
$16,000,000
51.32%
$7,741,844
$14,700,000
52.67%
-1.34%
6.07%
8.84%
Adirondack
$7,035,147
$13,900,000
50.61%
$6,748,333
$12,900,000
52.31%
-1.70%
4.25%
7.75%
Silver Meteor
$42,400,000
$84,800,000
50.00%
$42,600,000
$80,500,000
52.92%
-2.92%
-0.47%
5.34%
New Haven-Springfield
$11,944,482
$24,300,000
49.15%
$11,723,569
$24,300,000
48.25%
0.91%
1.88%
0.00%
Lake Shore Limited
$35,200,000
$71,900,000
48.96%
$35,000,000
$67,600,000
51.78%
-2.82%
0.57%
6.36%
City of New Orleans
$23,200,000
$47,600,000
48.74%
$22,000,000
$43,300,000
50.81%
-2.07%
5.45%
9.93%
Capitol Limited
$23,500,000
$48,600,000
48.35%
$22,600,000
$46,600,000
48.50%
-0.14%
3.98%
4.29%
Texas Eagle
$30,000,000
$62,200,000
48.23%
$28,500,000
$62,500,000
45.60%
2.63%
5.26%
-0.48%
Pere Marquette
$3,152,828
$6,900,000
45.69%
$3,276,210
$6,000,000
54.60%
-8.91%
-3.77%
15.00%
Crescent
$34,500,000
$78,100,000
44.17%
$34,900,000
$76,500,000
45.62%
-1.45%
-1.15%
2.09%
San Joaquins
$39,401,591
$89,500,000
44.02%
$38,661,536
$86,500,000
44.70%
-0.67%
1.91%
3.47%
Coast Starlight
$47,700,000
$109,600,000
43.52%
$45,300,000
$100,500,000
45.07%
-1.55%
5.30%
9.05%
Ethan Allen Express
$2,825,134
$6,500,000
43.46%
$2,829,307
$5,100,000
55.48%
-12.01%
-0.15%
27.45%
Vermonter
$5,029,712
$11,600,000
43.36%
$4,761,018
$10,700,000
44.50%
-1.14%
5.64%
8.41%
California Zephyr
$55,700,000
$129,900,000
42.88%
$53,200,000
$123,500,000
43.08%
-0.20%
4.70%
5.18%
Illini
$9,562,149
$22,400,000
42.69%
$9,258,647
$21,400,000
43.26%
-0.58%
3.28%
4.67%
Blue Water
$6,228,730
$14,600,000
42.66%
$6,094,659
$15,300,000
39.83%
2.83%
2.20%
-4.58%
Silver Star
$37,800,000
$89,100,000
42.42%
$38,700,000
$83,900,000
46.13%
-3.70%
-2.33%
6.20%
Cascades
$29,269,205
$69,400,000
42.17%
$30,886,455
$68,600,000
45.02%
-2.85%
-5.24%
1.17%
Southwest Chief
$49,100,000
$118,200,000
41.54%
$48,200,000
$114,900,000
41.95%
-0.41%
1.87%
2.87%
Wolverines
$19,398,853
$46,700,000
41.54%
$17,704,897
$38,500,000
45.99%
-4.45%
9.57%
21.30%
Chicago-St. Louis
$16,382,439
$39,700,000
41.27%
$13,353,833
$39,600,000
33.72%
7.54%
22.68%
0.25%
Piedmont
$3,325,948
$8,300,000
40.07%
$3,077,031
$7,200,000
42.74%
-2.66%
8.09%
15.28%
Capitol Corridor
$27,699,783
$71,500,000
38.74%
$27,927,540
$75,500,000
36.99%
1.75%
-0.82%
-5.30%
Kansas City-St. Louis
$5,617,913
$15,500,000
36.24%
$5,139,069
$15,800,000
32.53%
3.72%
9.32%
-1.90%
Illinois Zephyr
$5,788,619
$17,700,000
32.70%
$5,687,467
$17,500,000
32.50%
0.20%
1.78%
1.14%
Cardinal
$8,600,000
$26,600,000
32.33%
$8,400,000
$25,600,000
32.81%
-0.48%
2.38%
3.91%
Sunset Limited
$13,800,000
$54,700,000
25.23%
$13,000,000
$54,800,000
23.72%
1.51%
6.15%
-0.18%
Heartland Flyer
$2,022,956
$8,500,000
23.80%
$2,086,587
$9,200,000
22.68%
1.12%
-3.05%
-7.61%
Hoosier State
$892,553
$8,900,000
10.03%
$856,675
$4,700,000
18.23%
-8.20%
4.19%
89.36%


Special attention should be paid to Washington-Norfolk. Despite being in its inaugural year of service and being a single frequency with some fairly ugly timing (5am departure from Norfolk, 8:50pm return), it managed to make a profit to the tune of about $130,000.

It is a bit concerning that costs have increased so dramatically on some routes. The Hoosier State, for example, nearly doubled its expenses and I cannot think of a rational reason why. It's a short distance train that runs only four times a week, yet has a higher allocated cost now than the Heartland Flyer, Piedmont, Pere Marquette, Ethan Allen Express, or Virginia's Lynchburg and Norfolk lines. As for the other lines, I suspect that new allocations for the Northeast Corridor may be largely to blame. I'd like to say that Amtrak's taking possession of the former Norfolk Southern track in Michigan and upgrading to 110mph service is to blame for the Wolverine's increase, but that wouldn't explain the Blue Water's decrease or the lack of change for Illinois.


Change in expenses
Hoosier State
89.36%
Ethan Allen Express
27.45%
Wolverines
21.30%
Maple Leaf
16.20%
Piedmont
15.28%
Pere Marquette
15.00%
Palmetto
14.53%
Carolinian
11.71%
City of New Orleans
9.93%
Coast Starlight
9.05%
Downeaster
8.84%
Vermonter
8.41%
Empire Service
7.89%
Adirondack
7.75%
Washington-Lynchburg
7.41%
Lake Shore Limited
6.36%
Silver Star
6.20%
Silver Meteor
5.34%
California Zephyr
5.18%
Illini
4.67%
Pennsylvanian
4.49%
Capitol Limited
4.29%
Cardinal
3.91%
San Joaquins
3.47%
Southwest Chief
2.87%
Empire Builder
2.62%
Crescent
2.09%
Cascades
1.17%
Illinois Zephyr
1.14%
Chicago-St. Louis
0.25%
New Haven-Springfield
0.00%
Sunset Limited
-0.18%
Washington-Newport News
-0.32%
Texas Eagle
-0.48%
Keystone Service
-0.84%
Auto Train
-1.11%
Kansas City-St. Louis
-1.90%
Acela
-2.05%
Pacific Surfliner
-4.29%
Blue Water
-4.58%
Northeast Regional
-4.95%
Capitol Corridor
-5.30%
Hiawathas
-5.58%
Heartland Flyer
-7.61%
Washington-Norfolk


As before, an Excel version of the file is available for those who wish to see the original and the numbers are from Amtrak's September monthly performance report.

7 comments:

  1. Interesting that Chi-StL posted the highest gain in cost recovery. This shows the benefits of capital investment that boosts speeds. 110mph operation is only occurring on a short segment right now, but the reduced travel times combined with the hype and publicity has boosted ridership while holding costs steady.

    ReplyDelete
  2. It's not even worth analyzing these numbers. The reason it's not worth analyzing them is the overhead allocation. We know from other publications that the Palmetto is profitable on a direct-costs basis and the Lake Shore Limited is close to break-even, which means that there's more than $15 million in allocated overhead going to the Palmetto and something like $30 million in allocated overhead going to the LSL.

    The allocated overhead is nearly as large as the direct costs.

    The result is that none of this is meaningful. Small changes in overhead allocation cause massive swings in reported "profitability" but don't mean anything.

    "Fully allocated costs" numbers are USELESS from a business perspective. Amtrak, like all railroads, is a capital-intensive, high-overhead business. "Direct costs" or "Incremental profits" numbers would be useful. So would analyses of how existing capital could be leveraged to run more service with the same overhead. Would a second NY-Chicago frequency over the same tracks and the same stations improve Amtrak's bottom line? It probably would!

    By contrast, the "Fully allocated costs" numbers are nearly worthless for any business analytics. The "fully allocated costs" will never tell you when a second frequency will be profitable, because the overhead allocations will completely distort the picture.

    ReplyDelete
    Replies
    1. I'm afraid I have to disagree with you there. Remember, that allocated overhead involves anything that cannot be definitely restricted to a single train. This includes station costs. So while staff at San Juan Capistrano can be directly costed to the Pacific Surfliner, staff at Los Angeles (which additionally serves the Sunset Ltd, Southwest Chief, and Coast Starlight) cannot and must be allocated formulaically. This actually represents a weakness for trying to use direct costs as they represent an underestimate for most routes rather than an actual marginal cost. Fully allocated costs, while higher than marginal, remain important for looking at route performance in general. For instance, looking at the numbers derived from this (for a future post), I can see that overnight sleepers have a seat-mile cost 40-45% higher than an all coach day train.

      As for whether a second NY-Chicago frequency over the same tracks and stations would improve Amtrak's bottom line, that's dependent on rather a number of things and actually rather doubtful. Given the current average fare, (a measly 16.5¢ per passenger mile), a train with the same occupancy levels and composition would need a marginal cost under $30.79. Metrolink's marginal cost is $33.05 and Amtrak's is assuredly higher (the Surfliner was in the high-40s a few years ago).

      Delete
    2. Paul,
      I'm going to come down the middle between you and Nathanael. On the one hand, the fully-allocated numbers are open to all sorts of shenanigans involving administrative costs that would likely not be affected if a route were discontinued (certain overhead expenses at 50 Mass, for example). Allocating things like that tends to be open to abuse, and though a number of the swings seem to have at least some basis, you're right that in a few cases it feels more like Amtrak used a random number generator. The Hoosier State is particularly glaring because not only because it's 4x weekly, but also because it lacks food service. The only excuse I can think of for that is that there's a bad practice with the operating crews getting held over at one end, and even that feels insufficient. Upstate New York also seems to have gotten screwed as well.

      On the other hand, there are some costs that you rightly point out need to be considered (some station costs come to mind...but even there, taking a lot of the southern CA stations, those are shared with Metrolink/Coaster and so much of the cost probably shouldn't go to Amtrak...but "dumping" those costs to other tenants would indirectly screw the relative performance of LD trains). The problem is that the question of "can this cost be allocated" simply isn't being asked. There are costs that probably simply can't be traced in any meaningful manner and which, for the sake of any meaningful business decisions probably should be set aside.

      Now, there are two questions on a hypothetical second LSL that come to mind:
      -The first one is how much more you'd spend those allocated costs? Buffalo-Depew is a 24-hour station, for example, and there are a couple of other "high hour count" stations on the line between there and New York where the incremental cost of an additional train would be negligible...so the second train would probably "save" on the cost of the existing trains on that line.
      -The second one is what would the "network effect" be of that second train? Basically, if the second train had passable hours for Cleveland and Toledo (the current LSL isn't great in either place) and/or offered a sufficiently different time into New York (early enough to connect with the Star, for example), you'd probably add ridership there. You'd probably also induce some ridership on all existing trains on the route (and a lot of the lines with connections that would work with it)

      The reason I raise these points is that it's quite arguable that a second train might not directly cover its direct costs (per your analysis), but that induced ridership other routes would make up the difference...and that the indirect cost picture would be absurdly complicated to boot.

      -------------------------------

      One other thing I'd note: The VA situation is obviously complicated. WAS-NFK's figures are going to rise next year, since the train was only in operation for about 9.5 months last year. Additionally, the oddness in WAS-NPN's numbers comes from the fact that Amtrak didn't bother to fix the accounting between the routes, so that train is "billed" for five daily trains until 12/12/12 and four thereafter. It looks like that line was subject to the allocation shifting that a lot of other NEC-related routes seem to have gotten, but the effect was hidden by what "should" have been a 10-15% decline in costs (depending on overall costing trends).

      Delete
    3. As Anon notes, a second LSL on a different schedule wouldn't increase station costs hardly at all. Every staffed station on the route is already open 24 hours, with the possible exceptions of South Bend and Toledo. Syracuse is 24-hour, Albany is 24-hour, I believe Springfield MA is 24-hour, even Utica is 24-hour...

      There's a reason I picked the LSL an example; there aren't any plausible allocated costs which are actually going to increase by running a second frequency. No new stations, no new station employees, no new crewbases. Perhaps any maintenance costs for cars which can't be assigned to a specific route, but that's all.

      Economies of scale!

      And as Anon says, a second LSL at different times would likely more-than-double ridership due to better scheduling and network effects - second frequencies on half-decent routes usually do more-than-double ridership.

      By contrast, running a new train on a new route (a Pioneer or North Coast Hiawatha) would be a disaster in cost terms, as it *would* add a bunch of new allocated costs (new stations even).

      Delete
  3. I have reconsidered my claim that the numbers are completely useless, after realizing that on several different trains the allocated costs are quite close to the direct costs -- meaning a fairly standard percentage (100%) of the direct costs.

    This means anything with better than 50% "cost recovery" can be, very roughly, considered to be making a contribution to the bottom line (though not enough to cover its share of overhead).

    Anything in the 45%-50% range might be pushed into direct profitability with minor improvements; 5% is within annual variation. Of course, one must also consider whether a train is generating more than its share of overhead (as some of the cross-Rockies trains certainly may be); if that overhead can't be spread out over more trains (a second LSL might fill up, a second SWC won't) then the train may be bad for the bottom line anyway.

    Finally, the trains with less than 45% cost recovery need serious improvements or will require permanent subsidies (and not just for overhead).

    Most of these are corridor trains which are already getting state subsidies, and which clearly need to be faster or more reliably on-time.

    The LD trains in this category also have fairly obvious, well-defined problems, and to the extent possible those problems should be fixed before examining them further; if those problems can't be fixed, *then* it's time to pull the plug.
    (1) Sunset Limited and Cardinal simply must be daily. Daily or nothing. And the Sunset Limited needs to go to Phoenix, and the Cardinal needs to make decent time from Indianapolis to Chicago, and the Cardinal's "Buckingham Branch" problems need to be fixed.
    (2) SW Chief needs to go via Wichita and Amarillo. And cut a lot of padding out of the schedule, much of which is for the slow Raton Pass track.
    (3) Silver Star needs to either be reliable on time within Florida, or faster at getting from Florida to NY, or both -- it may need to be two different trains.
    (4) CZ needs to find a way to stop doing the brutally slow mountain crossings, regardless of scenery, needs to stop damaging Denver-Chicago timings with delays west of Denver, needs to go through the actual cities in Iowa, and needs better Salt Lake City service. If there were money for rolling stock and trackwork, I'd design a complete restructure: Glenwood Springs-Denver "Ski Train", Oakland-Reno "Extended Capitol Corridor", Denver-Omaha-Moline-Chicago "Denver Zephyr", and Oakland-Salt Lake-Wyoming-Denver "California Zephyr" -- the last being questionable.
    (5) The Coast Starlight needs, somehow, to be faster. Not sure how, over those mountains.
    (6) The Crescent needs to do something about the terribly weak south-of-Atlanta section. Faster, carrying fewer cars, something.

    The routes above my 45% cutoff mostly have fewer immediately obvious "fix this like this" problems, which is to be expected if my guesses about overhead are correct.

    ReplyDelete
    Replies
    1. Silver Star may not make sense to run when the Meteor is doing so much better. Functionally, Meteor takes two hours to serve Lakeland and Tampa. That's just not worth it, either ax the Star or have the two trains go over radically different alignments.

      We're spending money on two trains, the way it ought to go is one over Florida East Coast Railway (Jacksonville to Miami took 6 hours instead of 9, serves WAY more of the state), the other over the current Silver Star Route, and an experimental third over the old Seaboard/Palmetto service, via Waldo, Ocala, Wildwood and Dade Cityand making Gainesville* within reasonable distance** to Amtrak service.

      The way we've currently got it, with two trains, departing and arriving within an hour of each other, but one taking two hours longer to serve two more cities is absolutely asinine. Maybe there's some operational reason for that, but I don't know what it would be. If there could be three trains a day, with each running different hours (FEC has to be daylight, Silver Star could be run overnight so long as Orlando isn't too late, Palmetto would need to serve Tampa in daylight if Star didn't), that'd be ideal for Florida. The fact that Amtrak considered daily Chicago-Miami an absolute necessity at inception but doesn't find it necessary now is absolutely asinine to me.

      *My hometown, home of the flagship university of the state w/ 50,000 students.

      **RTS provides bus service to Gainesville Regional Airport currently, Waldo's amtrak station is 11 miles farther, rather than Amtrak running a thruway coach it could just pay for RTS to run one bus a day the extra 22 miles. Honestly either the city or University would probably foot the bill without Amtrak paying.

      Delete

Note: Only a member of this blog may post a comment.