I’ve had my disagreements with Randal O’Toole, a libertarian defender of suburban sprawl, but to his credit, he’s done the most convincing accounting of subsidies (well, accounting costs, at least) that I’ve seen yet. And though he normally concentrates on federal costs, his write-up of an American Bus Association report includes this paragraph about mass transit:
What about state and local subsidies? A first approximation of such subsidies can be found by subtracting expenses from revenues in National Transportation Statistics. The results suggest that total subsidies to air travel are tiny, subsidies to highways are large (but tiny per passenger mile), and subsidies to transit are in between (but much larger per passenger mile). National Transportation Statistics doesn’t have state and local subsidies to Amtrak or intercity buses, but I suspect the former are much larger than the latter.
All of this is probably true, but I’ve criticized the use of “per passenger miles” in the past (as had Michael Lewyn, unbeknownst to me at the time) on the basis that trips in areas served by mass transit can be shorter than trips made with in the suburbs and the exurbs with a car. I emailed Randal O’Toole and asked him what he thought of this argument, and as always, he was kind enough to send me his response:
You make a valid point. But it is most valid in regions where transit is concentrated in dense areas and jobs are concentrated in those dense areas. In post-automobile regions, such as San Jose, Phoenix, and Houston, neither of those conditions apply. The same is true in pre-auto regions that have undergone massive decentralization, such as Cleveland and St. Louis. Even in Chicago and San Francisco, jobs have decentralized to the point where dense downtowns hold only a small share of the region’s jobs.
Wendell Cox has some great maps somewhere on one of his web sites, or at least in his slide shows, showing the portion of urban areas such as Portland that can be reached from a random starting point within so many minutes by transit and by auto. Unless the starting point is downtown, transit reaches only a tiny fraction of the area reached by autos.
In sum, outside of Manhattan, I suspect your point is valid only for people starting downtown — that is, at the hub of our hub-and-spoke transit systems. I am sure a detailed analysis could be done to prove whether or not this is true.
First of all, I’m not sure Wendell’s point is all that relevant here – the question is whether miles (in other words, the “tiny fraction of the area reached by autos” that he mentions) is even a useful measurement when comparing transit-oriented cities and car-oriented sprawl. But the point in his first paragraph is undoubtedly true: most regions do not have very many jobs downtown, and transit works most effectively when everyone commutes downtown, within neighborhood, or somewhere in between the two. But is the paucity of jobs in cities’ dense urban cores a natural free market outcome, or is it the result of anti-density limits like zoning and parking minimums?
Furthermore, are transit lines not located in dense places (like most new lines out west) because there is no demand for density, or is it because they’re “zoned out” – that is, infill development is essentially no longer allowed? I know that every neighborhood near a transit line I’ve ever lived in has been completely built up to the zoning envelope. And the fact that most transit authorities allocate prime TOD property – the parcels directly adjacent to the station – to parking lots, completely ruling out any development whatsoever, can’t help matters. The problem of land socialistically allocated to parking near stations is most acute out west, where all new stations are surrounded by a sea of parking, but it also happens with commuter rail lines in the Northeast.