Last month the Pew Charitable Trust's SubsidyScope released a study of Amtrak subsidies. It made the news, and at the time I just shrugged it off as the kind of story that we'll have to put up with until (a) airline and private car subsidies come down enough to give trains the advantage or (b) Americans come to feel enough ownership of Amtrak to accept the subsidies the way they do airport and highway subsidies.
I also accepted the numbers, because I personally accept the need for Amtrak subsidies. Today, however, frequent commenter Bruce McF takes on the SubsidyScope numbers and shows that the work is a lot sloppier and less conclusive than I would have imagined from such a prestigious organization. No firm conclusions, other than that the subsidies are overstated.
The conclusion I got from Bruce's post is that depreciation is misallocated, not that subsidies are overstated.
ReplyDeleteBasically, the study makes the NEC looks somewhat better than it actually is and the other services look somewhat worse.
What Alon said - I focused on whether the SubsidyScope figures on profit/loss by route were of any use.
ReplyDeleteThe ones that are per passenger are just laughably bad. Even if all the depreciation allocated is rolling stock, allocating the same depreciation to each passenger on the Springfield/New Haven Shuttle or the Downeaster as to each passenger on the City of New Orleans or Southwest Chief is silly.
Per passenger numbers were their headliner - they only did per-passenger-mile as a supplementary.
And then in the per-passenger-mile figures, as Alon says, they clearly mis-allocated depreciation costs. I have not yet sorted out precisely which depreciation costs Amtrak omits from per-route figures that SubsidyScope includes, so I could not address how much of their "extra $24/passenger" is mis-allocated.
As to the overall averages - Amtrak claims that SubsidyScope's figures are inflated due to the nature of the lease-buyback that much of the rolling stock operates under. I did not address that claim, since I have not sorted it out yet. However, given the flaws I have already sorted out, I would not be surprised if SubsidyScope tripped up on the differences between cash flow and capital budgeting.
However, quite possibly a bigger element, which is just excluded by SubsidyScope's frame of reference, is tax recovery.
About 50% of Amtrak's budget is employment costs. If average income and other federal tax incidence on that is 20%, then 10% of Amtrak's budget comes back as taxes. Supposing that half of Amtrak's activity would have to close up shop without the Federal subsidy, then income taxes equaling 5% of Amtrak's budget would be lost, so cutting that 16% in funding would only recover a net 11%.
That's actually something that can be addressed with more precision, because Amtrak gives employment costs by main service line, so the lost wages and salaries from closing all the long distance routes can be given directly, and all the corridor routes that are not state subsidized can be given a rough estimate. And there is in the discipline a cost of unemployment literature to draw upon.
The depreciation of rolling stock should be allocated per car-km and locomotive-km, with a higher rate given to the Acela, which like all high-speed trains has shorter shelf life than low-speed trains and costs more per unit.
ReplyDeleteIt's not hard to do this. Each of the train orders has publicly available or FOIA-able costs and design shelf life.
The depreciation cost on track isn't harder. Again - it's all public, or FOIA-able. The trickier part is assigning shelf life to components. Curve eliminations in areas of Connecticut that will have to be bypassed should be reckoned as high-depreciation; constant-tension catenary is a 100-year asset.
FOI requests might be bit ambitious for a Sunday afternoon, but there's certainly information available that can be used to do a better job of allocating the depreciation by business line than SubsidyScope did, which was to not bother allocating the depreciation by business line and then act as if it did.
ReplyDelete