NARP Rebuttal of DOT Inspector General report on dining and sleeping cars

(Looking for brief information on this topic?  Read our news releases of July 14, July 26, and our Executive Summary)

September 19, 2005

To: Interested Capitol Hill Staff
From: Ross Capon, Executive Director, National Association of Railroad Passengers

This memo is prompted by the Association’s concern about the food and sleeper service language in the Senate appropriations bill.  To the extent that anyone who supports this language is sincerely trying to “help” Amtrak, they are probably relying on the DOT Inspector General’s report, “Analysis of Cost Savings on Amtrak’s Long-Distance Services.”  We think that the report is way off base.

We strongly oppose the dining car and sleeping car language in the fiscal 2006 appropriations bill passed on July 21 by the Senate Appropriations Committee.  We believe this language would fail just as miserably as did similar language in the early 1980s. If it is necessary to legislate on the appropriations bill in this area, we urge adoption of the report requirement contained in S.1516 as reported by the Committee on Commerce, Science and Transportation.
Richard Beadles—the last president of the Richmond, Fredericksburg and Potomac Railroad before CSX absorbed it, now a Richmond businessman and a NARP board member – said the following after reading the IG report:  “The OIG work product is reflective of staff which does not fully comprehend the business in which Amtrak operates…Some of us were around (yes, some of us are even guilty!!!) of the same sort of ‘cost-cutting cures’ during the period 1960-1971. Each cut cost more in revenue than it saved [emphasis added]. That’s how we got to Amtrak!!...I do agree, however, and I suspect Amtrak does also, that we’ve got to continue to look for better ways to deliver food and other services, at less cost.”
Tim Gillespie, Amtrak’s former Vice President—Government Affairs told me: “Legislating on food service costs was tried before [in the early 1980s], and it was a disaster. It hurt Amtrak’s bottom line and we had to undo it. (DOT Inspector General) Ken Mead ought to pay attention to his earlier comments that losses from the long-distance trains are ‘chump change’ compared to corridor investment needs. He’s focusing on the wrong problem.  Even if Amtrak could wipe out all the losses on long distance trains, it would still need almost $1 billion a year for the infrastructure costs and capital needs—that’s where Amtrak really needs help.”

The Association agrees that the OIG’s analysis is fatally flawed and that implementation would hurt, not help, Amtrak’s bottom line.  Ultimately, it would undermine support for the system, which depends on both revenues and political support from many income levels, and could not survive by serving only those people willing to sit up in coach seats overnight.  Elimination of sleeping-car passengers guarantees an immediate, huge loss of revenues, but the assumptions about costs that could be saved are highly speculative.  On its surface, the goal of continuing to run the existing number of train-miles makes obvious that significant costs will not change.
The Washington Times quoted me on the report thus:  “He [Mead] might just as well shut the trains down.  With the train, you take away the food service and you have made it impossible for the majority of the revenues to stay with you.  This is just a backdoor way of wrecking a $3 billion corporation.”
The Analysis is based on three false assumptions.

  1. Food & Beverage service is, or should be, a profit center.
  2. Only sleeping car passengers make long trips; coach passengers only make short trips.
  3. Elimination of dining, lounge, and checked baggage services would not affect coach revenues.

Because the assumptions are false, the conclusions and the recommendations should be rejected.
Amtrak Senior VP—Operations William Crosbie got it right when he testified on June 9 before the House Subcommittee on Railroads:  “Amtrak’s food and beverage service is a fundamental part of the service that we offer on board the majority of the trains that we operate…Its primary purpose is to enhance ticket sales and ridership, not serve as a profit center.”  There is merit to analyzing food service costs in relation to total passenger revenue, but not in relation to the additional revenue collected specifically from food and beverage sales.  On other modes, such as very-long-distance flights, where on-board time is comparable to that on the long-distance trains there generally are not separate charges for food, so from an accounting standpoint food service cost recovery is effectively zero. 
Substantial volumes of coach passengers travel long distances, accounting for a disproportionate share of coach revenues.  Amtrak data shows that, in Fiscal 2004, 1.7 million coach passengers (59% of coach passengers) on long-distance trains had trip segments of 400 miles or longer; of these, 827,000 were 800 miles or longer. Revenues associated with these passengers are quite significant—about $136 million or 78% of all coach revenues on the long-distance trains. Average trip length for all coach passengers is a misleading statistic because so few sleeping car passengers travel short distances.

(For trip-length-based breakdowns of long-distance passengers by volume and by revenues, click here. This data excludes Auto Train, since all passengers on that route ride endpoint-to-endpoint and excludes the New York-Chicago Three Rivers, which was discontinued during fiscal 2005.)
The need for lounge and dining service increases with trip length. This, coupled with the data on how many coach passengers make long trips, undermines the claim that elimination of such services would have little or no impact on coach revenues.
The need for baggage service is usually a function of length of stay away from home. That, again, is a factor independent of “coach vs. sleeper.”

When private railroads operated all-coach long-distance trains, they provided lounge, dining, checked baggage and attendant services.

Thus, the OIG’s plan for elimination of “amenities” not only would eliminate all sleeping-car revenues ($107 million in FY 2004, approximately 38% of revenues on these trains), it would significantly reduce coach revenues, with the lion’s share of the impact coming from the 78% of coach revenues associated with longer trips.

Two important caveats regarding the IG’s discussion of sleeping-car revenues.  The report did not mention that many sleeper fares are purchased months in advance, which makes those funds more valuable to Amtrak. Also, discounts such as AAA do not apply to room charges, only to the basic rail fare. The rail fare usually is much cheaper than the room charge.

What follows is a page-by-page commentary on the OIG’s analysis (OIG document page number in parentheses).
(2) OIG: The long-distance trains “are the only intercity passenger rail service in 23 of those states.”

NARP:  Although technically correct, the real number is 25, since the Oklahoma City-Fort Worth Heartland Flyer would not survive on a stand-alone basis, because nearly one third of its volume and revenue come from passengers connecting with the Texas Eagle.  Also, costs would rise for an isolated Heartland Flyer because it would bear the total cost of facilities at Fort Worth and require deadhead moves to rotate equipment to and from maintenance facilities, equipment moves now done economically using the Eagle.


(3) OIG: “Sleeper class differs from coach in that it provides a sleeping room…complete with turndown service, movie presentations and other entertainment (also available to coach passengers), and prepaid meals in the train’s dining car.”

NARP: This obscures two facts: Movies are also available to coach passengers on the Superliner trains—the movies are shown in the lounge car which is open to all passengers. On single-level trains, while movies were offered in sleeping-car rooms when the Viewliner sleepers were new, Amtrak is eliminating movies, citing costs of maintaining the in-room service and the growing tendency of passengers to watch movies they own on their laptops.

(4) There is no scientific way to allocate costs between coach and sleeper. Once the decision is made to run the train, any such allocation is arbitrary. Clearly, the OIG’s assignment of 100% of dining, lounge and checked baggage service costs to sleeping-car passengers is wrong.

The lounge cars (in which movies are shown) are heavily used by coach passengers. Indeed, in my own travels, I have used the lounge car more when riding coach than when in the sleeper. Sleeping car passengers pay for privacy, whereas coach passengers often use the lounge for a change of environment and the chance to socialize with someone other than their seatmate.

The very long trips that many coach passengers make challenges the IG’s assumption that one could eliminate diners and lounges with no impact on coach revenues.

The OIG’s tables which attempt to neatly assign precise subsidy levels to coach and sleeping-car passengers are meaningless. Such phrases as the following, based on the OIG’s incorrect conclusions, are wrong and should not become the basis for public policy: “disparity between the level of subsidies for coach class service and the level of subsidies for sleeper class” and “the cost of the sleeper class and other amenities is so expensive that the revenues pale in comparison.” These phrases are polemic and misleading because they obscure the high degree of subjectivity (and, in our opinion, inaccuracy) inherent in the OIG’s attempt to segment subsidy levels by class of service.

The right way to determine the economic impact of adding or subtracting services is at the margin—using marginal revenue and marginal cost. 
(4) OIG: “Amtrak must find ways to provide food service in a much more efficient manner to eliminate the need for Federal subsidies for food service.”

NARP: Most knowledgeable observers agree that food and beverage service cannot break even on a stand-alone basis. Frankly, adherence to that goal is tantamount to killing the service. However, there are opportunities to reduce costs by improving labor productivity, and to increase sales through better merchandising.

  • Witnesses at the June 9 House hearing on food service were in general agreement that the Amtrak/Gate Gourmet contract needs to be revised or dropped; even Gate Gourmet (which did not testify) was said to recognize this need. This covers food acquisition and delivering food to trains.
  • Because dining-car and lounge-car menu prices have risen so much in recent years (see “Increasing food prices” item below), see if there are opportunities to offer “early bird” and “late night” specials at lower prices to coach passengers.  Early-bird service, even at regular rates, would be particularly appreciated by families with small children, and by the elderly.  Coach passengers today get last crack at the reservations and often can’t eat easily.
  • Alternatively, consider adding to the regular menu a low-cost item, perhaps without dessert, that is popular and kid-friendly—spaghetti or meatloaf/mashed potatoes, with a vegetable.
  • To reduce on-board labor requirements while maintaining food quality, consider serving in dining cars the full meals that have in the past been served in Acela Express First Class.
  • Whenever ridership is below a certain threshold, consider dropping the reservation system and letting people eat when they want.
  • Review the possibility of once again requiring sleeping-car passengers to pay separately for their meals. This would run counter to management’s desire to minimize handling of cash on board. Moreover, new sleeping car fares would have to be adjusted mindful of the stiff increases in those fares in recent years even as amenities have been reduced. This approach would carry the danger of reducing labor productivity if meal sales fall too much.

(4) “Options for food service that could be tested on several long-distance routes include outsourcing, increasing food prices, having passengers obtain meals in stations during regular stops, distributing boxed meals that have been prepared off the train, and selling packaged food from carts on the trains.”
NARP: Below we explain why most of these options are non-starters.
Outsourcing is a possibility but not a panacea. Outsourcing commissary operations to Gate Gourmet did not deliver the predicted savings.  Also, Maine provides a $210,000 annual food subsidy for the Boston-Portland “Downeaster” trains (116 miles one-way). The menu is limited, satisfactory only for short trips, and sold by non-union workers who spend less time away from home than do Amtrak long-distance on-board workers. But, using the rough measure of food subsidy per passenger-mile, extending even this inadequate service and these unrealistic cost assumptions to the long-distance trains implies a food loss of about $27 million.
Increasing food prices would affect only coach passengers, since sleeping car fares include meals. Stiff menu-price increases have been implemented over the past few years, in both lounge and dining cars, provoking considerable criticism and arguably pressing the limits of what people of modest means (e.g., long-distance coach riders) will pay. Still more increases thus have a big danger of backfiring and lowering revenue. (See the discussion above on the page 4 item that begins “Amtrak must find ways…”)

Having “passengers obtain meals in stations during regular stops” is neither customer-friendly nor practical. Even if the stops were all spaced so that logical timetables could be constructed, placing trains at these stops at convenient times, key problems remain unaddressed:

(a) Passengers need food at other times;
(b) Some passengers have mobility issues;
(c) In extreme weather, many passengers likely would object to such a system;
(d) Trains do not always run on time—even a railroad with no capacity problems and perfect signal systems is going to have delays associated with freight derailments and mechanical problems;
(e) The sheer impracticality of expecting a station restaurant or deli to serve the required number of people without unduly delaying the train, which then adds to its operating costs and undoes the benefits David Gunn is trying to realize by eliminating mail and express freight. The operating economics would be grim for a station restaurant geared to handle large groups of people quickly, just twice a day. Likewise, the capital investments needed to produce adequate and reasonably consistent facilities at enough locations would be significant.

Amtrak already distributes “boxed meals that have been prepared off the train” in certain situations. See first page 5 item.
Amtrak has tried “selling packaged food from carts on the trains.” Indeed, NARP has recommended this on short runs where the alternative has been no food service at all. However, this is not practical for long-distance trains, and possibly not for high-volume corridor trains. OIG acknowledges that applicability is limited to packaged foods. In addition, the volume of food required on most runs would require “off-cart storage” space.

Employees using carts would have to be put up overnight, and some system would have to be set up to service their carts. Carts used on the Hiawatha’s and Pere Marquette can be plugged in during layovers but that’s used only to keep beverages cold (if it fails, the worst case is a warm Bud Light). It is unlikely that anything that requires refrigeration could be served, and that is a crippling limitation for long-distance travel.

There is also the inconvenience that the cart would create by occupying the aisle and blocking access through the train (or even between certain passengers and the nearest bathroom, as can happen on an airplane).
(5) OIG recommends “the initiation of pilot projects on select routes to test the effect of eliminating sleeper class and its associated amenities.”

NARP: We strongly oppose this suggestion, which reflects ignorance of moves that Amtrak has made in the past.  We dispute OIG’s apparent belief that everything on the train that is not a coach seat (or a bathroom?) is a sleeping-car “associated amenity.” Coach passengers need checked baggage service. Indeed, eliminating such is an invitation to safety problems as people try to bring everything into the coach.  As noted previously, coach passengers make heavy use of the lounge and even dining car services.

  • Amtrak ran its New York-Chicago “Three Rivers” with a single sleeper and scaled-down food service (box meals from lounge car with one employee) from April 1, 1999 until the sleeper was removed in November, 2004. The train itself was discontinued in March, 2005, but the five-month “sleeper-less” period probably is not a fair test since Amtrak had announced the train’s March discontinuance before the period began.)
  • Amtrak’s New York-Miami “Palmetto” ran without sleeping car from May 1, 2002 until the train’s overnight portion was discontinued in November, 2004. (The train now is a daylight service, running New York-Savannah only.)
  • Amtrak’s New York-Miami/Tampa “Silver Star” ran without dining car from January 1995 to February 1996, when the experiment was terminated.
  • For years, Amtrak has used boxed meals for sleeping car passengers for expediency on certain short segments. For example, the Empire Builder west of Spokane has two separate sections, Seattle and Portland. The Seattle section has the dining car (no lounge), the Portland section has the lounge (no diner), and the Portland sleeping-car passengers get boxed meals. Obviously, acceptability of boxed meals over a short segment does not assure acceptance of same for a much longer trip.


(5 and 9)  “Only a small number of riders (14% on all long-distance routes, and 9% excluding the Auto Train) of Amtrak’s long-distance trains take the routes end-to-end.”

NARP: The report misleads the reader into believing that only a small number of people take very long trips, thereby adding to the plausibility of talk about downgraded food service.

(9) “We also found that the average distance traveled is far less than the route length…The percentage of endpoint-to-endpoint riders on the California Zephyr was 4%.”

NARP: These are misleading and largely irrelevant statements, given the huge number of passengers making very long trips that do not coincide with the endpoints of Amtrak routes.

  • In FY 2004, passengers riding 800 miles or more on one train totaled 1.06 million (908,848 in coach, 151,102 in sleeper). [Using a threshold of 400 miles, the total is 2.20 million, again, excluding Auto Train and Three Rivers.]
  • A passenger riding from Cumberland to Flagstaff does not count as an “end-to-end” rider, yet his 2,320-mile trip is longer than an end-to-end trip on either of the routes involved (Capitol Ltd. 764 miles; Southwest Chief 2,256 miles). He is counted as a 629-mile Capitol Ltd. rider and a 1,691-mile Southwest Chief rider).
  • Similarly, passengers riding Naperville, IL, to Fullerton, CA—or even shorter segments do not count as “end-to-end” but nonetheless are making very long trips with the same amenity needs as “end-to-end riders.”

In short, the number of passengers who literally ride endpoint-to-endpoint has no significance because so many other riders also are making very long trips.  OIG’s emphasis on end-to-end passengers leaves one wondering whether the report’s authors simply don’t know the system, or are trying to exaggerate the number of riders whose amenity needs are very limited, to make the “eliminate-amenities” proposal sound more reasonable.
(8) “Across all long-distance routes, total revenue increased only about 1.5% from 1996 to 2004 in dollars adjusted for inflation.”

NARP: This is largely a result of shrinkage in the number and capacity of trains operated. During this period, Amtrak retired its Heritage sleeping cars and coaches, due to escalating maintenance costs. No new cars have been added to the long-distance fleet since 1996.


(9) “…the high costs of providing sleeper services and their share of associated amenities means that sleeper class passengers account for a disproportionate share of the Federal subsidies necessary to cover costs on these routes.”

NARP: Again, it is incorrect to assign to sleeping car passengers all amenities except for coach seats and bathrooms.


(10) “By providing just basic coach service, typical train sizes will decrease from two locomotives pulling 9 to as many as 13 cars to one locomotive pulling 3 to 5 coach cars.”
(13) “With much less weight to pull, only one locomotive would be needed for the train.”

NARP: When set up properly, the second unit is used primarily for accelerating away from stations and climbing grades. Thus eliminating the second unit does not cut locomotive costs in half.  Moreover, some long distance trains already have just one locomotive. On some other routes, the host railroad probably would insist on a second locomotive for reliability even if this would mean far more horsepower than train weight would dictate.


(10) “This change in long-distance rail service requires adopting a shift in thinking about rail passenger service akin to the major airlines’ moves to eliminate meals and other amenities in order to reduce their operating losses.”

NARP: Amtrak appears to be thinking hard about how to improve efficiency on these trains, but the airline analogy is not appropriate since so many travelers spend far more time on the train than most airline passengers spend on their flights, and most airlines still offer premium classes. As noted earlier, very long flights still offer meals at no extra charge.


(11) The OIG’s calculations “assume that…no coach passengers would abandon Amtrak if they no longer had access to amenities such as full-service dining cars, lounge cars, and checked baggage service. The net operating cost savings will be lower if any current coach passengers do not find Amtrak an attractive travel choice after those amenities are eliminated.”

NARP: The need for checked baggage service probably is a function of the length of time away from home, not coach vs. sleeper.

In addition, OIG has not taken into account passengers who connect between one train on which they ride sleeper and another train on which they ride coach, another reason why elimination of “amenities” would depress coach revenues.  Example: A passenger who wants to travel from Detroit to Denver and will only travel by sleeping car from Chicago to Denver (overnight trip).  If sleeping accommodations are not available, the passenger will choose another mode of transportation: costing Amtrak not just the revenue from Chicago to Denver, but the coach revenue from Detroit to Chicago as well.