It's completely invalid to pretend that none of the fixed costs of operating sleeping and dining car service is attributable to the passengers therein. If dining car cost cutting resulted in a significant decrease in sleeping car demand, Amtrak could pull sleepers offline and lay off porters. That would, in fact, save Amtrak money. In contrast, if sleeping car demand did not fall, then the cost cutting itself will save Amtrak money.
This is, fundamentally, what it means to say that longer distance sleeping car passengers receive a bigger subsidy (which they do).
The case for Amtrak's existence depends on the role the trains have in providing basic point to point transportation. And that function doesn't cost all that much, comparatively. Indeed, in the longer term, cost cutting could allow Amtrak to expand its long distance network.
Meanwhile sleeping car passengers who like sit down dining service not only cost Amtrak a lot of money, but really don't provide any compelling justification for subsidized transportation.
In the long term, the former model is sustainable. It will require subsidies, but it is sustainable. The latter model is not. Over time, Amtrak, as a matter of budgetary reality, will find ways to feed its long distance passengers more cheaply. And railfans will actually swallow their objections and continue to ride, because there is no alternative other than the private land cruise model which is far more expensive and far slower.
Enjoy the dining cars (and the sleepers, but that's longer term) while you can. Long term, that's where the cost savings are.
Eh, we're on the topic...may as well run with it. So, here's what I run into:
1) The diners are, always have been, and always will be something of a loss leader because of the staffing and capital requirements involved in operating them. However, it is worth noting that the language Amtrak used when discussing the LSL's diner situation last year seemed almost flabbergasted that coach passengers were grabbing at least breakfast or lunch in the diner, and making up something like half of the business overall.
2) With that said, since the fixed costs (operating the car and so forth) are a large enough share of things, there are ways to shrink the marginal losses. For example, assume that we carry out an exercise on a train...keep roughly the current size diner and feed as many tables with it. How many tables of four could you feed at a reasonable pace? I'm guessing at least 12 (48 seats), and possibly 14-16 (56/64 seats)...which is a far cry from the 8 tables (32 seats) that you get stuck with on occasion when a 40-seat dining car (since the staff often takes up two tables for paperwork). I would also point out that the dining car staff can often take take-out orders even when the tables are full, pointing to sub-optimal utilization of staff.
3) Moving on to sleeping car passengers, my understanding is that (with a few "bad route" exceptions...Cardinal and Sunset Limited, I'm looking at you) sleeper passengers pay their full weight as far as the cost of operating the train, and in fact usually wind up effectively cross-subsidizing the coach folks.
4) Following up on this, there's an old report called "Amtrak 90" that got linked to on here today. I won't get into the details, but if you can add capacity to an existing train, you tend to shrink the per-passenger losses (and generally the overall losses) by a noticeable amount. Look at the cost recovery on the Empire Builder, the Coast Starlight, or the Auto Train (hint: Which trains haven't gotten reviewed yet because they're in the upper third of LD trains? Those three plus the SW Chief and the CONO). I forget the exact numbers, but I think it was something like adding two sleepers and a coach to a Superliner train added 18% to operating costs but 45% to operating revenue.
What's the takeaway? No, the diners aren't going to make money...but you can push them further and close those per-passenger losses noticeably, and if you can do this plus add more sleeper space, you can dramatically improve train performance. A mid-bucket Silver Service train that adds a sleeper, filled at an average of $300/roomette [assuming 10 are filled] and $600/bedroom [assuming 3 are filled] (trust me, this tends to be on the low side once you factor in the coach fares for double occupancy), you add just over $3.5 million to the revenue of that train per year.
At this point, it's not cost-cutting that Amtrak needs...it's capacity expansion. If the former can permit the other, that is great...but right now, what Amtrak needs above all else is more space on many trains. Between 2005 and 2011, Amtrak posted the following gains on the Silvers:
Star:
2005: 405/train
2011: 581/train (+43.5%)
Meteor:
2005: 395/train
2011: 512/train (+29.6%)
What does this tell me? On the one hand, the trains are increasingly popular; on the other hand, that without new equipment, capacity is a problem. Without capacity, it is hard to push revenue up but /so/ much (since fare increases, however ambitious, can only take so much of the load). I'll say it again: Amtrak needs more cars to run on the existing trains so it can reduce losses.


