There is an interesting long thread that has been running on railroad.net on the subject of sleepers, their revenues and their direct and indirect costs, as an incidental subject, discussed in excruciating detail and unfortunately not unequivocally positive for sleeper service. Indeed one of the most vocal participants is a proponent of just worrying more about Coach trains, and he does make a very cogent case for it too. Thought some of you might want to at least be aware of that line of argument, since in this board we seem to mostly have arguments about minutia on this subject rather than some of the basic premises.
Anyway here you go:
http://www.railroad.net/forums/viewtopic.php?f=46&t=136456
I've read it, and I think the fellow who believes sleeper service is less profitable than coach is cooking the numbers to get his preferred result -- all our available numbers are a bit unclear, but that fellow is biasing every one of them to make sleepers look bad and to make coaches look good, including bizarre claims that sleepers cost more to maintain than coaches (in reality, they cost less to maintain because they don't get used by so many people). He's been completely unable to justify the claim that replacing a sleeper with a coach on the same route will collect the same revenue in general (though I would believe it on certain routes with exceptionally poor "through" demand and exceptionally strong "local" demand).
The differential results of different segments of trains makes it really easy to come up with misleading averages, which is a trap everyone seems to fall into.
It's pretty clear that if you take
(a) a good-enough-to-get-lots-of-customers train route (so, not Atlanta-New Orleans),
(b) which is already running coaches and a diner,
© and which has demand for overnight service,
you can improve the bottom line by adding sleepers.
Yes, that's a rather specific claim I'm making.
Even the fellow who's making the full-on attack against sleepers surrendered and admitted that sleepers are probably profitable on the LSL. Contrary to his claims, they're profitable on the Silver Service too, where Florida-NY traffic is large and isn't going to ride coach.
There are problem segments, where the train simply isn't competitive with driving (Atlanta-New Orleans, Grand Junction-Reno). Here, adding any car may lose you money, sleeper or coach. As far as I am concerned something should be done about these segments; limping along is not a good option. I will discuss these problem segments below, since this has turned into a full-blown essay.
The situation with the dining cars adds great confusion to most people's calculations. The dining cars are simply a loss leader; you need them for people who are riding too long, because the alternative is "get out at Albany for a meal break", and that suppresses ticket revenue by a large (though very hard to calculate) amount. I assert that the dining cars are necessary even without sleeping cars -- if the trip runs across two mealtimes, the cafe car can't even stock enough food for everyone, and people start wanting a sit-down meal -- so I assume the existence of the dining car when figuring whether sleepers are profitable.
It's very hard to figure out how much the dining cars *should* be costing right now because they are currently antique cars (in the East) operated with archaic paperwork-heavy practices (nationwide). But I will say that dining cars are probably costing a lot more than they should.
After Point of Sale is implemented in the dining cars, and the Heritage cars are replaced, it may be possible to get a better picture of the "true" dining car cost structure going forward, and maybe Boardman is right and they can be profitable. I could certainly imagine that if they were open most of the day and right up to the end of the trip, rather than closing for inventory periodically, it would make a big difference.
In some sense sleepers and diners are a sideshow. Amtrak *should* have them on routes of appropriate runtime length, as they will improve revenue by more than costs and the bottom line.
But the fundamental issue is, is the route operation any good to start with? Again, I will discuss this below, because it turned into an essay.
Sleepers, coaches, diners, all are mere details in comparison to the basics of running daily, on time, at a speed competitive with driving, between cities of meaningful population -- which increases the ridership and the prices Amtrak can charge by very large amounts. This *should be* possible on most (not all) of the so-called "long-distance" routes, but is not currently achieved on any of them.
An all-coach Lake Shore Limited with the current unreliability problems would just cost more money, and a "no-diner" Lake Shore Limited would just cost more money. The same applies to the Silver Service. The focus should be on eliminating the unreliability by one means or another.
BTW, after you eliminate the "lemon socialism" and "lemon capitalism" crowd in Congress do you really have enough votes to actually pass anything that we would favor?
Quite a cynical viewpoint, and I'm rather cynical. I think the privatize-the-profits, socialize-the-losses people are actually a small minority in Congress (although there are a lot of such people attempting to *bribe* or *mislead* Congressmen).
The group you have to target for votes is the "pork barrel", "log rolling", "something for me, something for you", "The Erie Canal will be good for local business" people -- who are subtly different, and a very large group. The privatize-the-profits, socialize-the-losses people will be pushing dishonest propaganda at this group (watch for words like "boondoggle"), and the most reliable response when talking to the "pork barrel" types is "look at the boost for local business/tourism in your district, look at how many of the voters of your district use this".
Now, it can be hard to sell amenities to a Congressman that way, so the point should be that they actually improve the bottom line by increasing revenue and ridership. (If a particular amenity doesn't, well, I can't really support that amenity either.)
BTW, Boardman has already presented one slideset with a bar graph that Nathaniel had posted somewhere which clearly shows that if the 6 LD trains that lose most money above rails and after removing all shared costs were eliminated that would leave only the east of Chicago LD system standing. Everything west of Chicago will be gone.
Here starts the essay.
There are several types of problem segments.
- The first is the segment with low population where all forms of transportation are marginal -- such as the High Line in Montana and North Dakota. This is actually the least problematic, as the train will get a large percentage of the traffic.
- The second is the segment with low population where the roads are significantly faster, and it would cost a completely unreasonable amount to speed the railway up -- such as Denver to Salt Lake via the mountains, or Salt Lake to Reno. Those mountain/desert/prairie crossings with low populations and good roads are mostly like this.
- The third is the segment where the roads are significantly faster than the railway even if the railway is running on time and well-maintained, but the population is large enough that it would be *worth* speeding the railway up -- such as New Orleans to Atlanta. In these cases it has been very hard to find the funds, but the funds *ought* to be found.
- The fourth is the segment where the railway, even on its existing route and right-of-way, ought to be about as fast as the roads, but deferred maintenance, abusive dispatching by the Class Is, failure of the 14th St. yard to do its job, etc. etc. has prevented it from running on time. This is true, to some extent, of most Amtrak routes on freight tracks. These are the most *anger-inducing* segments for me, because these perfectly-good routes are basically being sabotaged.
The Texas Eagle is actually a dramatic example of problem segment type 4. There's plenty of on-line population. There's nothing wrong with the route of the tracks: it's only a little twistier than the parallel Interstate. However, it is much slower on every single segment. I know why from Ft. Worth to Dallas, from Alton to St. Louis, and from Joliet to Chicago, and in all three cases it's just stupid stuff which could be fixed pretty easily if anyone cared. I don't know why in the other segments (this route is too slow on nearly every segment), but I'm guessing it's more of the same "easy to fix but nobody bothers".
- The fifth is three-a-week service. This is guaranteed to perform poorly, and is exceptionally poor at providing network connectivity. It's not even worth considering other problems with the Sunset Limited or Cardinal until the three-a-week problem is fixed. These do, however, provide some network connectivity, as the ridership in the through cars on the Texas Eagle shows.
Due to the fourth phenomenon, all long-distance routes are performing worse than they should -- and so are most of the state sponsored routes. But let us discuss the first three sorts of "problem segments".
At current speeds, Indianapolis-Chicago and Atlanta-New Orleans are problem segments of type 3 -- the routes are simply too slow to be competitive, but funding for a better route *could* make them competitive. Denver-Salt Lake City through the mountains is a problem segment of type 2. Albuquerque to Topeka via Raton Pass and the middle of nowhere is also a problem segment of type 2.
Now, here's the thing: you can't just delete the problem segments -- of any of the types -- because if you do, you lose network connectivity and ridership drops across the board (this was the greatest error in the Carter cuts). You have to figure out how to improve them or provide better substitutes for them. Railroads are all about network effects and economies of scale.
The problem of finding funding for problem segments which provide important network connectivity has been the most difficult problem in the history of Amtrak. A few have gotten funded: in terms of those which connect to trains on both ends, New York to Boston (which was definitely a problem segment of type 3 when Amtrak started), St. Louis to Kansas City, and Charlotte-Raleigh. Also a number of state-sponsored "branch lines" which connect only at one end. But most have not gotten funded.
Each problem segment needs its own custom solution. This is partly what the Performance Improvement Plans were for, but they couldn't recommend major capital improvements, which is the solution for several of the problem segments of types 3, and they can do very little about the sabotage which creates problem segments of type 4, for which the only reliable solution has been to buy the tracks. They *did* recommend daily service for all trains, addressing type 5 problems, but we still haven't got it!
The problem segments of type 1-3 have repeatedly been tacked onto other segments. Which makes sense as these segments can't stand on their own; and in some ways, this avoids making them targets. But it has a bad side effect: it makes the other (better) segments into targets. The bad segments are used, through misleading and selective averaging, to make routes with high potential look bad.
VIA, which is trying to kill the Ocean, pulled this trick by pointing to the low ridership from Halifax and ignoring the high ridership from Moncton. (Then they made it three-a-week in a further attempt to kill it.)
For a US example:
I am quite certain that a standalone Denver Zephyr would look pretty good financially (18 hour train, 15 hour drive, quite competitive) -- but the result would be that it would become obvious how much money it cost to run the California-Denver train. But cutting the California-Denver train would lose connecting service from the ski areas to the East and from Reno to the west, reducing ridership systemwide.
If you somehow managed to keep ski areas-Denver and Reno-Oakland funded as separate corridor trains (which has been made harder by PRIIA, but not impossible), you'd still be losing the connecting revenue from the people taking really long trips, and you'd have to add extra expenses to run a Denver maintenance shop (and to find train parking at Reno, and at the ski areas), and you'd lose the votes from Utah. (If there are any pro-Amtrak votes from Utah.)
Salt Lake - Reno is a problem segment of type 2, but useful for connectivity. With its low population and the amount of time it takes it should be run overnight. Unfortunately, Salt Lake - Denver via Wyoming is a problem segment of type 2 or 3, which you would also want to run overnight at its current speeds -- while Salt Lake - Denver via the Moffat Tunnel Route is a problem segment of type 2. I guess Amtrak selected the current scheme as the best of a bad lot of options.
Running more trains per day along any of these routes would allow for better scheduling for each of the separate markets, and could multiply revenues by more than costs as the increased frequencies make the train fit into more people's schedules. In some cases (problem segments of types 1 and 2) there is likely not enough demand. But in cases of type 3 (and 4 and 5!) there is enough potential demand, and more trains per day would make the capital costs (which are necessary to improve the "problem segments") more justifiable.
Amenities simply must be provided at the level which keeps people buying high-priced tickets. Sleepers are simply something you run on overnight trips. Diners are something you run on trips which run across multiple mealtimes. Cafes are something you run on trips which run more than about, say, 2 hours. Lounges are something you run on trips which are long enough for people to get antsy, maybe more than about 6 hours. If you don't run any of these on the appropriate type of trip, you bleed customers and people pay much less for tickets. As long as you're running any of them, tart them up and make them look nice -- it costs very little and improves word of mouth reports a lot.
But the real issue is the routes and schedules themselves, and critically, the on-time performance.
There is a strong case to be made that rail service over very long (in terms of hours) routes is not really viable, coach or sleeper. Obviously, you can have overlapping corridors, making a route much longer than the typical trip taken on it, but at some point the cascading delays and odd calling hours start deteriorating the value of the train for any of the sub-corridors.
Consider, as a thought experiment, this alternative to the California Zephyr: a train running on the same schedule from Chicago to Denver, and a train running on the same schedule from Denver to Oakland, with a "guaranteed" connection at Denver.... but where both trains left on time and people on misconnecting trains were accomodated in the hotel at Denver Union Station. This way, delays coming from San Francisco would not impact the passengers going from Denver to Chicago, only the passengers coming from west of Denver. About half the passengers on each train wouldn't even notice the difference, according to the PIP (half the train turns over in Denver).
(This was a lot easier and cheaper to do when railroads owned their own hotels at the stations. Frankly, Amtrak should lease out the upper floors of Chicago Union Station to a hotel with an agreement that the hotel will house displaced Amtrak passengers for a flat rate.)
The other half of the passengers on the CZ would notice the change of trains at Denver. Those who are travelling from east-of-Denver to the ski areas probably wouldn't be too bothered by the change of train (though more bothered by an overnight delay). This implies setting up a separate Ski Train on a schedule which can pick up most misconnections.
The through traffic apart from "ski traffic" is fairly small, and biases towards the sleepers for obvious reasons. These people are travelling long distances slowly, and are going to be less bothered by the delay and more bothered by changing rooms. So if you want to accomodate that, you run a through sleeper, and if the train misconnects, they stay in their rooms and it gets attached to the next day's train. (Requires some spare sleepers, obviously, and even spare OBS staff.)
This is beginning to sound like a very traditional operation, isn't it? Coach trains on relatively short routes with reliable schedules, with a few sleepers being transferred from one coach train to the next making some extra money. There's a reason some of these practices developed.
The dismemberment of the railway system in the US prior to Amtrak and during the early days of Amtrak has made a lot of this stuff seem very hard to do, due to the very low number of trains per day running and the extreme difficulty of restarting a simple local route. These facts are obscuring the true nature of "added amenity" service.
The problem with the long-distance trains is not the amenities (which actually improve their performance), it's the delays, the uncompetitive schedules, the low frequencies, and the routings through low-population areas.
In technical (as opposed to political) terms, the easiest route to prove my thesis with is the Lake Shore Limited -- fix up both ends of the route, run it on time, keep adding cars and frequencies, and you'll see; the amenities are profitable, it's the the underlying schedule and timekeeping problems which were unprofitable. Other routes which could be used to prove this would be a restored Broadway Limited (with the west end fixed up), a Denver Zephyr at 90 mph through the major cities of Iowa, or the Silver Meteor rerouted onto a restored 110 mph S line between Raleigh and Richmond and running on time in Florida. Which route is politically easiest to prove this with is a matter of the current prevailing winds.
Or if you want to prove my thesis the bad way, extend the Palmetto to Florida without sleepers or diners and watch it perform worse than the Silver Meteor. But I don't suggest it.
I wish there were Congressional hearings on Amtrak's lack of on-time performance. That would actually have been *useful*.