Given the numbers from AlanB and others, it is quite clear that Amtrak would be better off eliminating coach and making the trains all sleepers.
Why must the solution be some ridiculous extreme?
Just to illustrate the logical conclusion of this kind of proposal.
The conclusion is only "logical" to those who misunderstand the facts.
The only subsidy that a sleeper passenger gets is that which a coach passenger gets, namely the rail fare subsidy.
why not eliminate this subsidy?
I think whoever spends $3,556 to travel SPK-PDX-LAX (the OP's example) isn't getting much of a subsidy. I'm all for maximizing sleeper revenue, but it seems perverse to set prices so high that rooms run empty, just to eliminate a notional subsidy.
The problem with this entire discussion is that the discussion is centered around the idea of a subsidy "per passenger."
No such subsidy exists.
The train costs X to run. Much of that is spent whether you have nobody on the train or 400 people on the train.
Then there are step costs that are fixed across a certain range, but may go up or down if that range is exceeded (adding an extra coach, for example).
There are other costs associated with certain service levels (running a dining car vs. a lounge-only train).
To offset those costs, you need revenues. Most of those revenues will come from the passengers who buy tickets. There are very few cases where the marginal cost of transporting an individual passenger exceeds the revenue from that passenger.
It is also absolutely ridiculous to suggest that any train in the system would do better financially if it carried fewer passengers.
You really can't say you are going to "eliminate the subsidy" for a certain passenger or a certain group of passengers. The subsidy is for the train, not the passenger. The passengers offset the subsidy with ticket revenues. For those who want to "eliminate the subsidy," what they are really saying is "eliminate the train." If the fares were to increase to the level required (under current ridership) to cover the full cost of operation, the effect would be a reduction in ridership and revenue that would greatly exceed whatever is gained from the fare increase. (I will note the irony that there are folks who complain about Amtrak fares, particularly sleeper fares, being "too high" and then, at the same time, say we should not subsidize sleeper travel, ignoring the painfully obvious reality that lowering the fares on a service that is already frequently sold out increases the overall loss for the service, thereby increasing the subsidy required to keep the train running...simple arithmetic, people).
The figure of "subsidy per passenger" (or "subsidy per passenger-mile") or any other hybrid of total current loss and current ridership numbers is not a valid tool for planning. The only thing it is useful for is cheap shots by politicians and anyone else who wants to campaign against passenger rail. If you really want to make an honest financial analysis of the service with the intent of finding ways to improve it, then you must understand that the subsidy per user number is a byproduct of the financial performance, and not an actual input into the equation.
