Saturday, November 23, 2013

Amtrak Food and Beverage

Back on Halloween, Amtrak’s Office of Inspector General released a new report on the state of Amtrak’s food and beverage service. Unsurprisingly, it continues to lose money, to the tune of $72 million in Fiscal Year 2012. There are several issues, however, which it overlooked or otherwise did not appear to notice which I thought worthy of comment.

The first is that the Acela’s food and beverage profitability, and by extension that of the Northeast Corridor as a collective whole, does appear to be a case of cooking the books. Unlike the complimentary food on the long distance trains, which is accounted for according to the menu price, Acela first class meals are attributed according to a calculation based on hotel meals in four cities which Acela serves. In May of 2011, this formula was adjusted with the result of a significant increase in revenues attributed to food and beverage.

Put plainly, this sort of formula is complete crap. Even if there is not a true menu price due to the meals not being sold to other passengers, a notional menu price should still be created, if solely for bookkeeping reasons, based on the cost of food and a normal and customary markup. The prices of meals at hotels near Amtrak stations has absolutely nothing to do with the cost or appropriate price of the food which Amtrak is serving. It may serve as an appropriate guide to the food which Amtrak should serve or the general price range of its offerings, but it cannot be used in isolation of the actual food itself.

Next we come to the Auto Train, the single most pointless train in all of America. The sole justification for the Auto Train is that it makes money; it provides no useful transportation service such as one might argue with the other long distance trains. Predictably, it loses tens of millions of dollars each year. I find it somewhat surprising that rail advocates who argue passionately for more long distance sleepers do not advocate the breaking up of the Auto Train and the distribution of its sleepers amongst other routes which might find more use from them; perhaps they fear the logical conclusion of an opportunity use argument against long distance equipment in general.

That aside, the Inspector General is swallowing camels while straining gnats when it comes to the Auto Train, marking only an objection to the Auto Train, and three long distance routes, spending $480,000 on complimentary wine and cheese tastings. Quite frankly, this is a pittance and, if one insists on investigating this, a better question would be how much does Amtrak make in revenue from sales of wine and cheese as a result of the tasting. Instead, the Inspector General should have questioned just why Amtrak spends quite so very much on food and beverages for passengers of the Auto Train.

For every single passenger aboard the Auto Train, Amtrak spends a total of $52.89 in food and beverages. That’s not the menu price, that’s not the price including labor, that’s strictly the commissary cost per passenger. To put the problem another way, for every three dollars which Amtrak collected in ticket revenue from passengers aboard the Auto Train, one dollar was spent on food and beverage service. Now, the Auto Train is naturally going to have a higher amount of spending than any other train; in view of the high fares which are paid in order to transport their vehicles, every passenger is given complimentary food, not merely the sleeper passengers. Perhaps this is a reasonable business decision by Amtrak to drum up additional business. What is not a reasonable business decision by Amtrak is that this extends to complimentary alcohol as well. Alcohol is one of the most profitable things there is in the food industry with markups of four to five times the actual cost to the restaurant. It should never be given away just for free like that, but only in such a fashion as to increase sales otherwise (such as a wine and cheese tasting which then offers bottles of wine for sale).

Let’s return to that commissary cost, $52.89 per passenger. That’s the cost of a single dinner and a continental breakfast and the food cost of a continental breakfast is extraordinarily low; a typical price including labor and profit margin is generally under $10 per person. If this represents a normal per person figure for sleeper passengers on other trains, there are some profoundly disturbing implications for the question of whether sleepers actually result in a marginal contribution to train revenues.

I grant that use of a contracted commissary may result in higher food costs than direct purchase of raw ingredients and as such, Amtrak can expect a higher ratio of food costs than the 33% of sales that restaurants normally go for. However, this ratio is never less than 42.9% (Silver Star) and exceeds 100% on six trains: the Auto Train, Sunset Limited, Crescent, Southwest Chief, Lake Shore Limited, and the Texas Eagle. That does not account for the cost of labor, which is tremendously high. A waiter, or any other Amtrak on board service employee, receives a wage of $25.54 per hour and has a total cost, including benefits and employer paid taxes, of $41.19 per hour according to the Office of Inspector General. This is far in excess of the $8 to $10 an hour that a normal waiter may receive here in the state of California, with higher levels of paid benefits and employer paid taxes as well for Amtrak employees, and in both cases the employees receive tips.

I do not begrudge anyone a living wage and even with positions that are normally heavily tipped,  I am strongly in favor of a rate of pay which allows them an affordable living without a reliance on tips. But Amtrak’s labor costs are insanely out of line with anything that could ever be profitable and are significantly higher than what is necessary for a living wage. Generally I oppose comparisons to other industries, but in this case I think it is a valid one: Is Amtrak on board service truly worth being recompensed, before tips, at the same rate as nursing?

But, though the long distance dining car is doomed to make direct losses, perhaps it makes an indirect profit. Perhaps the dining car, and the free food that it provides for sleeper passengers results in sufficient additional sleeper revenue as to overshadow the losses. This may, perhaps, be true on some trains. It is most certainly not true on the Sunset Limited, Cardinal, and Crescent, where the cost of providing food and beverage service exceeds the total revenue from all sleeper passengers; in the case of the Sunset Limited, even with the revenue transfer from the sleepers, food and beverage losses exceed total sleeper passenger revenue.

Let’s assume that coach passengers on other trains, by and large, follow a similar pattern to the Palmetto, with the same average revenue and costs ($3.99 and $5.09). This will underestimate things on both accounts slightly of course, since some coach passengers do purchase meals (at a loss for Amtrak) from the diner, but useful for a first run through. Under this assumption, even the best performing train, the Silver Meteor, has food and beverage expenses amounting to 33.7% of total sleeper revenue, an average amount which comes up to 66.6% of total sleeper revenue, and the always terrible Sunset Limited coming up to 150% of sleeper revenue. The vaunted sleeper fares, which Amtrak admits are not set with consideration of the cost of food and beverage service in mind, very quickly disappear once food and beverage service is brought into play.

Let’s look at it another way. Each car incurs a certain amount of costs per mile for fuel (.03 gallons according to Table 3.4) and maintenance. As a per mile charge, this necessarily results in a higher charge for the Western trains, which travel up to 2,438 miles per run, and lower ones for the eastern trains and tri-weekly Cardinal and Sunset Limited. In absolute terms, this number ranges from a daily train low of $489,256 on the Capitol Limited to a high of $1,529,237 on the California Zephyr. As a percentage of sleeper revenues, this generally runs the gamut of 5-10%, for an unstaffed dining car with no food in it, with the notable exception of the Cardinal, where it amounts to 21.64%, rather easily explained by the fact that the Cardinal possesses only a single sleeper. Of course an unstaffed dining car is rather pointless, so let’s add the costs of staffing to it. Admittedly, most trains include a separate lounge, so we’ll discount those trains by the labor cost per mile of the Palmetto, which is coach only with a dinette (88¢ per mile) and the Coast Starlight twice to remove the Pacific Parlour Car. This isn’t a perfect solution of course; wages are paid by the hour, not by the mile, and there are pre and post trip paid duties, but it should get us into a reasonable ballpark. The result? Just under half of Amtrak’s sleeper revenue is dedicated to dining cars and that before they’ve even paid for any food. For the daily trains excluding the Auto Train, the average sleeper fare is only $71.91 higher than the average coach fare once such figures are deducted; a fare that must still account for the actual cost of food, the dedicated sleeper attendants, and the opportunity cost of using a lower capacity sleeper car instead of a coach car.

Is this an unfair assessment for the sleeper passengers? Unfortunately no. The dining car exists solely for the purpose of sleeper car passengers. Passengers on the Palmetto find themselves perfectly satisfied with a dinette despite traveling the same distance as coach passengers on the Coast Starlight. Yet the Coast Starlight finds it necessary to add not merely a dining car, but also a second lounge, the Pacific Parlour Car, dedicated for the sleeper passengers. What revenue is spent by coach passengers in the full dining cars, over and above the cost of food, serves to subsidize the sleeper passengers. Even then, even if every single dollar of revenue attributed to the diner came from cash, at no point does the revenue exceed the labor hour cost plus food cost (it exceeds labor cost by about a nickel in June).

Until the 1980s, food in the diner car was strictly a paid for extra, no matter whether one was a coach passenger or a sleeper passenger. It is time for Amtrak to either return to that tradition or to adjust sleeper fares in order to recover the full cost of providing complementary meal service to the diner passengers. Advocates for the long distance trains like to crow about how high the occupancy factors are for these trains (a feat accomplished by having a small number of seats); they should have no problem with fares increasing in order to recover costs until such time as total revenue actually begins to drop. An additional benefit of returning to the prior tradition of all passengers paying at the diner is that it should decrease demand somewhat, allowing for more cars to be served with a single diner.

Amtrak should also divest itself of the Auto Train. As I mentioned before, it serves no justifiable purpose except to make money, which it manifestly fails to do. It is possible that the equipment on the Auto Train (8 Superliner sleepers, 4 deluxe sleepers, 8 Superliner coaches, 6 diners, and 4 lounges) may allow for better cost recovery on the other trains. This sort of self-justification is something which every Amtrak supported route should be regularly subjected to; routes should never be kept running with abysmally poor ridership simply because they currently exist and nor should they do so in the hopes that there may be a future service improvement, especially since the capital cost of equipment should be accounted for. Yes, subsidies may be necessary for certain social goals, but they should still be given out on the basis of the biggest bang for the buck.

This is a question that is going to especially trouble the Western long distance trains as the Superliner equipment wears out and new bilevels are available thanks to the corridor purchase order by California and certain other states. Are these trains worth a capital investment of seven hundred million, likely over a billion dollars with new locomotives, and an ongoing annual subsidy of three hundred and fifty million dollars? Would the nation be better served instead by using the money for other programs (such as Essential Air Service) or by funding alternate capital improvements, such as an extension of the San Joaquins to Redding or Chicago to Omaha several times daily? These are questions that will be asked by Congress in the coming years and they do not bode well for the long distance trains until and unless they are able to remarkably improve their performance by, among other things, getting rid of losses for food and beverage service.

An Excel copy of the spreadsheets created to work on this is available here.

8 comments:

  1. I have noted in the past that the subsidy per passenger of EAS is far in excess of the subsidy per passenger of Amtrak. So I would argue that while Amtrak needs to make its accounting less opaque (and separate into at least three different operating divisions), and that the commissary needs to be revisited, EAS monies outside of AK and HI would actually be better off invested in Amtrak, and then reduced once the operating needs for such service is "discovered".

    ReplyDelete
  2. "This may, perhaps, be true on some trains. It is most certainly not true on the Sunset Limited, Cardinal, and Crescent,...."

    The Sunset Limited and the Cardinal are essentially non-functional trains due to the three-a-week schedule. It is worth ignoring them in pretty much all comparisons. They will never be able to charge appropriate prices as long as they run three-a-week. If they ever run daily, we might be able to compare them with other routes. I can only call them expensive placeholders for daily service; and clearly they only exist because Amtrak thinks it will be harder to get service restored to the line if they are removed (probably a correct assessment).

    The Crescent is suffering a problem due to the south-of-Atlanta section, which runs so empty that Amtrak doesn't mind cancelling it for four days a week for two months every year. Amtrak is currently trying to find a new station site at Atlanta where Amtrak can drop half the train.

    This south-of-Atlanta section is priced very cheaply, runs very slowly, contains multiple mealtimes, and is probably dragging the numbers way down. This is almost certainly why the food numbers look bad on the Crescent. And, unlike the Cardinal, the south-of-Atlanta section doesn't attract political support.

    These are the three where the food seems to be losing money -- and as usual, it's not because of the food. It's not even because of the sleepers. It's because there's something defective with the route as a whole.

    We know why the three-a-weeks are still three-a-week -- terrible freight railroad obstinacy.

    Why does the Crescent still run south of Atlanta?
    (1) The route has potential if anyone ever put money into it
    (2) Connectivity to New Orleans adds unknown numbers of passengers to the system as a whole. Network connectivity matters.

    I still don't think cutting anything makes any sense. It is unfortunately clear from past experience that cutting long distance trains doesn't get you anything at all in the way of replacement.

    If a state is actually willing to pay for a valuable corridor service and the only thing holding it up is lack of equipment, then maybe it would make sense to steal the equipment from a long-distance service. But that's not what's been happening.

    "Is this an unfair assessment for the sleeper passengers? Unfortunately no. The dining car exists solely for the purpose of sleeper car passengers. "

    It is an extremely unfair assessment for the LSL! 50% of dining car patronage on the Lake Shore Limited is from coach passengers -- this is from the PIP. I realize the LSL is somewhat unusual in this regard, but whenever someone makes a silly attack on "long distance trains" in general, they end up implicitly attacking the LSL, which is usually doing much better.

    ReplyDelete
  3. (Part II.)
    As for the Auto Train -- The Auto Train breaks even on direct costs. Which includes the food. Cutting it would save $0.00. There isn't a better train to allocate the Superliner sleepers or diner to. (The better-performing trains are Viewliner trains; the other Superliner trains perform worse.) Don't be misled by the meaningless "fully allocated" costs; incremental costs are what matters when you're talking about cutting or adding a service. (And Amtrak releases them very rarely.) I'd love it if there were a better train to allocate this equipment to, but there isn't, financially speaking...


    Now, you need a diner for overnight trains, sleeper or no sleeper. Removal of the diner has been shown in the past to crash coach and sleeper ridership both. If you don't want to run overnight trains, you need to run the trains faster, because people (and politicians) want to have trains from NY to Florida and NY to Chicago and Denver to Chicago. I'll get back to this in a minute...


    "This is a question that is going to especially trouble the Western long distance trains as the Superliner equipment wears out and new bilevels are available thanks to the corridor purchase order by California and certain other states."
    The trains which will most be troubled are the ones with no constituency. These are also, not coincidentally, the worst performing per-passenger-mile and per-passenger. And the slowest. And the ones with the least investment.

    The worst is the Sunset Limited. Frankly I'm surprised it's survived this long -- it has next-to-no political support. If it doesn't go daily I don't think it has a chance. Even if it does, if it doesn't make it back to Phoenix it will still be obviously at risk. In its current form, I won't miss it -- three-a-week is frankly unusable.

    So, there's another theme which I haven't quite gotten to here. A bunch of the long-distance routes are problematic in their current state, but have a lot of *potential*. If there are state and local governments actively working to speed up the tracks they run on and improve the location of the stations they stop at (adding connections) and reroute the trains for faster service, these routes are simply going to get better, and so I say we keep them. Routes running on the NEC, or Empire Corridor, or under Englewood Flyover, or from Chicago to St. Louis, or across Tower 55, or into Denver Union Station, will perform better as the improvements are built.

    But some routes, and portions of routes, seem to simply be stagnating -- and that means they're always going to be this bad. There seem to be absolutely no improvements in the pipeline for the Sunset Limited.

    The contrast between the routes with investment and those without is already pretty striking -- and it will become even more striking in the future.
    So at some point Amtrak will have to give up on the worst routes, if nobody is willing to put any capital funds into them. Amtrak will need to put its operating funds where its capital funds are. North Dakota ponied up money for the Devil's Lake line -- but Texas hasn't put in a cent on the Sunset Limited route.

    I think Boardman actually understands this. His 'campaigning' around the nation for the long-distance trains has really been of the form of "Are you, locally, willing to put in capital funding?"

    ReplyDelete
  4. Considering the auto train actually is doing the best that can be done with its equipment (see above), I think it should be tweaked, improved, and emulated.
    They did do a recent tweak: letting people pay more to get their cars off first.
    http://www.progressiverailroading.com/amtrak/article/Amtrak-adds-capacity-on-Auto-Train--38889

    I also don't think that using less fuel and preventing wear and tear on vehicles, removing them from highways, is a non-useful service to provide

    Offering the wines (and cheeses) from the tasting for purchase, and doing something (anything!) about the obscene costs of food would make it profit on direct costs. The train already stops in SC for refuelling- Allowing non-vehicle passengers to board or alight in North Charleston (already on the route) would work to bring in passengers that are cheaper to accommodate (no automobiles, no sleepers, no free meals). Sunrail serves sanford to give non-auto passengers a connection everywhere from there to Orlando, Kissimmee. I don't know what there is between Lorton and DC, but a motorcoach might be profitable to run.

    I've written up a long post of what I'd like to see amtrak do in Florida. I'll post it here and in another one of your entries as a comment. Looking forward to your reply.

    ReplyDelete
  5. Florida's Amtrak service is absolutely ridiculous, compared to any similarly populated region of the country.

    The silver services carry alignments and schedules identical to, often worse than, the ones developed in the 1930s, when the entire state of Florida had about as many people as the Bronx, and steamboat and other water-based transportation was still commonplace. They have speeds today inferior to the ones they did then. Florida has nearly as many people than the STATE of New York now, a greater number soon, and our highways are better maintained than nearly any others in the nation (it's terrible driving anywhere else)- but they get congested, because they're the only viable option.

    Amtrak has lagged so far behind in serving the Florida of this century, that a freight railroad is introducing passenger service here.

    Now, that's not a bad thing for the future of passenger rail in Florida. There actually are a lot of excellent factors going for Florida:
    The new Sun Rail commuter rail system is the first in this whole part of the country- but the connection and feeder between orlando and the I-4 corridor isn't the biggest part of that-
    CSX got a bundle of money from the moving of freight traffic off of their A-line, and used it in double-tracking, grade separating, and signalling their S-line to build a rail superhighway from the Port of Tampa to the Port of Jacksonville and. That's in anticipation of the Panama Canal expansion project, because Tampa is much more well-suited to intermodal container transshipment than Miami or Port Everglades. But that capacity isn't running out soon, and it's for time-sensitive freight, so it's far more conductive to passenger carrying, plus the track is new and/or well maintained.

    All Aboard Florida (i.e. the Florida East Coast Railroad) has made massive improvements between Miami and Coca, and is laying new rapid rail track from Orlando International Airport to their mainline. A next phase is most likely going to Jacksonville.

    Amtrak's most profitable and best-loved train outside of the NEC is based in Florida (The Auto Train).

    A huge portion of our population is potential rail users: Nearly half of our population is either under 24 (as you know, millennials are driving less, or don't even have licenses- but love to travel) or over 65 (They may drive to the store, but many aren't up for going 100+ miles behind the wheel, and they LOVE bus tours- more of them would take Amtrak if greater service were available)

    Intermodal transportation and Florida's tourism sector were made for each other- It's now possible to fly into the state, take a train, and catch your cruiseship at the port, without ever using a car. We get a lot of european travellers- the more train options there are, the more of them will use the train.
    (continued, #1)

    ReplyDelete
  6. Given all this, it's asinine that Amtrak continues to give florida less service than Florida had 80 years ago.

    Here's what I'd like to see for changes in existing lines:
    Re-route the Palmetto to go from DC to Tampa, extended from savannah to Jacksonville, then via CSX's S-line. That would re-activate service at Ocala and for Gainesville, which means over 65,000 college students and 600,000 people overall who currently have only 2+hour bus rides for thruway service receiving Amtrak service. Redcoach is making a killing offering service from UF to Tampa. It's good local ridership and it opens that cooridor.

    Re-route the Silver Meteor over the FEC right of way. Service along the A1A/I-95 cooridor from jacksonville to miami- that 5+ hour drive is an easy one to compete with, and FEC is already outfitting it for ideal passenger use. Amtrak currently takes nearly 11 hours to make that trip- FEC main line could take it down significantly- if the stretch between West Palm Beach and Cocoa is indicative, quite possibly under 4 hours.

    But that's not the most important part- Florida's population density is concentrated along the our coast. Nearly 10 million live in the counties along the atlantic between jacksonville and Miami. The current alignment grossly underserves that population (and a huge tourism sector). They'd use rail if it were a reasonably competitive, or at least available, choice.

    Silver Star: This route may just need to be rethought entirely.
    Amtrak probably can't to this in their current sitatuion:
    (What's most needed is conncecting, frequent, cooridor service betwen Tampa/St Pete and Orlando International Airport to fill the gap left by the other options- running along the SunRail tracks toward the coast would be ideal- especially if a connection could be made betwen Deland (existing) and Daytona (a new station along the FEC right of way). Go from there to jacksonville and then along the current meteor alignment. Florida HSR would have been a real asset in making any amtrak service between tampa and orlando unecessary, but we have what we have)

    Auto-Train: I can't offer much to improve the most profitable and best-loved service, but here's two possibilities:
    1. Let non-car passengers buy tickets. Thruway could be provided at Lorton (sunrail covers orlando area transfers) or passengers could arrange their own travel.
    It might not be a lot, but they'd be gravy anyway (pay more than they cost operations)
    2. Move the refuelling from Florence to Charleston (technically North Charleston) and allow non-car passengers to board or exit: but from a specifically designated car to make the service indestinguishable to current riders. Again, there may not be a lot, but Gravy. The new Charleston station will be co-located with the Airport, so there are possibilities there as well.

    (Continued, #2)

    ReplyDelete
  7. I'd also like to see a major re-imagining of Sunset Limited:
    Sunset was truncated out of Florida and has been the worst performing train in the country ever since. This is a mistake that needs undone.
    Furthermore, given the massive length of the journey, some operational changes could really improve things.
    Here's my re-imagining of the line:
    The only long-distance, non-NEC interstate line to perform excellently is the Auto-Train. That should be a lesson.
    But I don't think that limited stops are the Auto-Train's special sauce- I think it's vehicle carriage (and maybe the wine tasting, though probably not).
    So here's the new itinerary:
    Origin: Sanford. Automobiles only. Typical wine and cheese, leave the station in the afternoon (specifics depend on track/freight realities).
    Route: Connect to CSX's S-line, then turning west to lake city and onward
    This bypasses Jacksonville, but unless the "amtrak probably can't do this" improvements are done, jacksonville is a rather inconvenient to serve form Sanford, and in my scenario, would be a well-served stop anyway. This route connects UCF, UF, FSU, basically 150,000 college students, by one rail service, for the first time in Florida's history. From there, gulf coast, new orleans, houston, and then STOP in San Antonio. (this would be around lunchtime the next day). All passengers detrain for lunch and a layover in San Antonio. Vehicles can be removed at this point, or be put on the train.
    Another wine/chesse tasting, because why not?
    Between 6:00 and 8:00, the train leaves San Antonio. It goes the current route from there, although ideally the non-service for phoenix would be fixed, but barring that, thruway services could connect passengers there. Terminus: New auto-train station in San Bernardino, California.
    This is a comparable drive between Los Angeles and San Diego, and MetroLink can take the non-auto passengers to a variety of other destinations from here. Arrival would be sometime in the afternoon or evening (again, depending on track and freight realities).

    The layover helps in ensuring that even if the first section of the journey is behind schedule, that the delays won't easily get the train stuck waiting behind slow-moving freight. It also gives another vehicle destination (in the booming texas triangle) and point of origin.

    Driving across the country is enough of a pain point that this could be important- I originally envisioned a (non-amtrak) service from El Paso to Beaumont with intermediate vehicle on/off-loading at san antonio, just because driving through texas is such an awful part of driving across the country.

    There is no currently way to get your vehicle across country available and well-known to most people other than driving for basically (optimistically) 3 straight days. With a service like this, the advantage in rolling resistance of steel-on-steel and the lesser wind resistance of running in consist provide an environmental benefit, and the service upholds rail service which otherwise wouldn't be viable on passenger service alone (e.g. the gulf coast stations)

    (Continued, #3)

    ReplyDelete
  8. Finally, here's what I'd like to see for a new line, and my conclusion:
    Renewed service from Chicago, via the midwest, tennessee, and Atlanta, then to savannah, jacksonville, the CSX S-line, and terminating in Tampa, or reversing from Tampa and continuing to Miami (if we're interested in maintaining the odd horseshoe in the middle of Florida that currently exists, this route is how).
    North of I-4, Florida's more integrated with Georgia than South Florida. No direct rail service is inexcusable. It's slightly asinine that Floridians on Amtrak can only go to Chicago if they go through D.C. first, so this fixes that as well.

    It's time that Amtrak started serving Florida like a state with nearly 20,000,000 people and the nation's most important tourism sector.
    Florida's more conductive to railway expansion than many realize.

    These routes would cover the vast majority of Florida's population- Add thruway between Naple/SW Florida and Miami+Tampa, and that's functionally the entire state's population served by Amtrak. It's possible, and it's the right move.

    (Final, #4)

    ReplyDelete