While total net subsidy may be of interest to policymakers, additional insight may be gained by normalizing the subsidy; that is, by dividing it by an indicator of the magnitude of the transportation activity being subsidized. This makes it easier to compare subsidies directly among modes that vary dramatically in their extent and utilization. Normalization may be accomplished in several ways depending on the intended use.
Passenger-miles are the most basic measure of passenger transportation mode usage. They take into account both the number of passengers using a mode and the number of miles each passenger travels on the mode. It is likely that most forms of user benefits, and even many non-user benefits, are proportionate to mileage, and thus it can be used as a proxy for total benefits. It is comparable across modes.
However it is still an imperfect proxy because across transportation markets there are generally diminishing returns to trip length i.e. that revenues per mile are lower for longer distance trips, other things being equal. This would suggest that the same would hold for the benefits of a transportation mode relevant to evaluating a subsidy for that mode. Comparing modes that have dramatically different average trip lengths, subsidy per passenger-mile may overstate the subsidy for modes with short trip lengths and understate subsidy for modes with long trip lengths. Differences in circuity among modes will also impact results, since measured trip length will differ among modes for the same origin-destination pair.
There are also problems with the availability of data for passenger-miles for some modes.
The number of passengers can serve as a proxy for passenger-miles in cases where the number of miles per passenger (or per trip) are roughly similar for different passengers. The data are also generally more readily available. Because this measure does not reflect mileage, however, it does not distinguish between a short distance trip of little traveler value, with few social costs, and a long distance trip of larger traveler value and larger social costs. While subsidy per passenger-mile overstates the subsidy for modes with short trip lengths, subsidy per passenger understates the subsidy for such modes. The other major problem with this measure is that for some modes, like transit, only unlinked trips are generally available. Data on unlinked trips treat each leg of the trip as a separate trip, so that a single round trip may be represented in the data by four or five unlinked trips. This may overstate ridership for such modes.
3. Seat Miles
Seat miles do not measure actual transportation usage, but instead measure transportation availability. The argument for this approach is that government subsidies to carriers essentially purchase availability, and the utilization of that availability is determined by private carrier pricing and other market conditions beyond the governments control. While this measure cannot be used now, due to data limitations among the modes, it could be a helpful additional metric when the data become available. It would still be limited as a basis for broad comparisons, since most government expenditures on transportation do not go to carriers, but to infrastructure, and its not clear what meaning subsidy per seat mile would have for personal vehicles, for example.
4. Route Miles
Another way of looking at subsidies could be relative to route miles. Some subsidies are used to add incremental infrastructure, and arguably what is being purchased in that instance is not so much usage as access, and access is measured by the extent of routes over which service is available. There are inherent limitations to this approach, especially for modes like air and bus where competing companies run parallel routes on the same infrastructure (i.e., do two companies on a route count as one route or two.) Vehicle miles could represent another approach to measuring access.
All of the mileage-based metrics assume that transportation between two points is the goal of the activity. Some forms of transportation, such as general aviation and boating, are heavily used for recreational purposes, where the objective is to enjoy the transportation activity and then return to the starting point. In these cases, passenger hours might be more suitable for normalization than mileage-based metrics.
While passenger-miles are used in this analysis, future work could explore the implications and data issues of using passengers, seat-miles, and other measures. This study uses passenger-miles because of their numerous strengths, as indicated above.