Wednesday, August 22, 2007
Internalizing Automobile Externalities: Mileage Charges or Fuel Taxes?
by Tom Bozzo
Gordon sarcastically notes that the 10 cent/mile estimate of the external costs is much lower than the 90 cents/mile he assumes in assessing the net costs of rail-based transit, which he regards as little more than a jobs program in most places. Wev. (Road building, being in most cases heavily subsidized by general tax revenues, may appear much the same.) Anyway, Parry et al. convert the 10 cents a mile into $2.10/gallon, of which only 40 cents is recovered by average U.S. gas taxes. Permanently add $1.70 to the price of gas and the result won't be a chorus of complaints that the U.S. offers too many mass transit options. Calculation of escape velocity for the collapsed remnant of the light truck market is left as an exercise for the reader.
Parry et al. are really jazzed about the idea of recovering the substantial accident-related costs via "pay as you drive" insurance, which is "emerging at the state level." Accident costs are strongly driven by miles traveled, but traditional insurance premiums are not. They cite a study that figures the premium for the average driver at 6 cents/mile. Actual premiums would adjust for individual- and vehicle-related risk factors.
The amount isn't totally unreasonable. As early-middle-aged Midwestern parents driving safe cars with clean driving records [**], and excluding coverage for damage to our own cars, we pay a bit more than 3 cents, which is the result of around 2 cents for the more heavily utilized car and 5.5 cents for the less-driven one.
What bugs me is the implicit alliance of the nanny state and the nanny corporation. Parry &c. note that "advances in electronic metering technology" make it "feasible to charge motorists on a per mile basis." Now, it's my understanding that automobiles have long been equipped with "odometers" that record vehicle mileage pretty accurately, and the data are even collected from time to time (e.g., when cars change hands). So what they're really talking about is liberating data that the cars already collect at much higher reading frequencies versus the current system. This doesn't necessarily require advances in electronic metering, but suppose it does.
Such a system would involve far more extensive data sniffing infrastructure than existing electronic toll-collection systems, since neither EZ-Pass-style transponders — i.e., not linked to the vehicle's computers — nor monitoring of a limited set of entry points would suffice. Electronic data collection costs would be a particular challenge for insurers, who have considerably lower customer densities than traditional meter-reading firms such as electric and gas utilities and (unlike utilities) can't count on the transponder being in any particular place at a given time. The bottom line is that government, probably via big-business contractors, would end up doing the data collection and sharing the results with interested parties.
Both government and insurers have interests in how, as well as how much, people drive; I'd expect agitation to mandate reporting of vehicle speed among other possible parameters. Plus, the systems will collect data on where cars were at various times. This may not be a big deal for a proto-panopticon country like the U.K. [***], but there remains at least the popular delusion in the U.S. that it's nobody's business where you legally go. All of this monitoring would be both expensive and have serious privacy implications.
So what of fuel taxes, which Parry et al. say may arguably have seen their "externality rationale" come and go? Well, they're strongly driven by vehicle mileage, for one thing. Vehicle fuel efficiency is related to vehicle mass and other characteristics that drive accident-related external costs — you want to get into an accident with a Honda Civic or a Ford Expedition? The entire fee-collection apparatus already exists, is visited regularly by all drivers, and doesn't care about when or where you drive. What's not to like?
Granted, increasing motoring taxes is not a political walk-in-the-park, but that's a function of the status quo. If the question is how to be charged the $1.70/gallon (or mileage-based equivalent), maybe we're better off with the devil we know and spending money on railroad track instead of roadway antennas.
[*] Ian Parry, Margaret Walls, and Winston Harrington, "Automobile Externalities and Policies," JEL vol. 45, 373-399.
[**] I.e., Americans who face relatively cheap car insurance rates.
[***] Though I gather that there is occasional civil disobedience.
Ken (with a h/t to Peter Gordon) pointed me to a recent Journal of Economic Literature survey [*] on the external costs of motoring — i.e., costs not directly borne by motorists, including those related to pollution, congestion, accidents, and oil dependency — and possible policies to address them. A main conclusion is that most of the marginal external costs are mileage-related rather than directly fuel consumption-related, so it may be more efficient to institute charges based on the miles driven instead of by fuel taxes. As I explain below, this is not so obvious.
Gordon sarcastically notes that the 10 cent/mile estimate of the external costs is much lower than the 90 cents/mile he assumes in assessing the net costs of rail-based transit, which he regards as little more than a jobs program in most places. Wev. (Road building, being in most cases heavily subsidized by general tax revenues, may appear much the same.) Anyway, Parry et al. convert the 10 cents a mile into $2.10/gallon, of which only 40 cents is recovered by average U.S. gas taxes. Permanently add $1.70 to the price of gas and the result won't be a chorus of complaints that the U.S. offers too many mass transit options. Calculation of escape velocity for the collapsed remnant of the light truck market is left as an exercise for the reader.
Parry et al. are really jazzed about the idea of recovering the substantial accident-related costs via "pay as you drive" insurance, which is "emerging at the state level." Accident costs are strongly driven by miles traveled, but traditional insurance premiums are not. They cite a study that figures the premium for the average driver at 6 cents/mile. Actual premiums would adjust for individual- and vehicle-related risk factors.
The amount isn't totally unreasonable. As early-middle-aged Midwestern parents driving safe cars with clean driving records [**], and excluding coverage for damage to our own cars, we pay a bit more than 3 cents, which is the result of around 2 cents for the more heavily utilized car and 5.5 cents for the less-driven one.
What bugs me is the implicit alliance of the nanny state and the nanny corporation. Parry &c. note that "advances in electronic metering technology" make it "feasible to charge motorists on a per mile basis." Now, it's my understanding that automobiles have long been equipped with "odometers" that record vehicle mileage pretty accurately, and the data are even collected from time to time (e.g., when cars change hands). So what they're really talking about is liberating data that the cars already collect at much higher reading frequencies versus the current system. This doesn't necessarily require advances in electronic metering, but suppose it does.
Such a system would involve far more extensive data sniffing infrastructure than existing electronic toll-collection systems, since neither EZ-Pass-style transponders — i.e., not linked to the vehicle's computers — nor monitoring of a limited set of entry points would suffice. Electronic data collection costs would be a particular challenge for insurers, who have considerably lower customer densities than traditional meter-reading firms such as electric and gas utilities and (unlike utilities) can't count on the transponder being in any particular place at a given time. The bottom line is that government, probably via big-business contractors, would end up doing the data collection and sharing the results with interested parties.
Both government and insurers have interests in how, as well as how much, people drive; I'd expect agitation to mandate reporting of vehicle speed among other possible parameters. Plus, the systems will collect data on where cars were at various times. This may not be a big deal for a proto-panopticon country like the U.K. [***], but there remains at least the popular delusion in the U.S. that it's nobody's business where you legally go. All of this monitoring would be both expensive and have serious privacy implications.
So what of fuel taxes, which Parry et al. say may arguably have seen their "externality rationale" come and go? Well, they're strongly driven by vehicle mileage, for one thing. Vehicle fuel efficiency is related to vehicle mass and other characteristics that drive accident-related external costs — you want to get into an accident with a Honda Civic or a Ford Expedition? The entire fee-collection apparatus already exists, is visited regularly by all drivers, and doesn't care about when or where you drive. What's not to like?
Granted, increasing motoring taxes is not a political walk-in-the-park, but that's a function of the status quo. If the question is how to be charged the $1.70/gallon (or mileage-based equivalent), maybe we're better off with the devil we know and spending money on railroad track instead of roadway antennas.
[*] Ian Parry, Margaret Walls, and Winston Harrington, "Automobile Externalities and Policies," JEL vol. 45, 373-399.
[**] I.e., Americans who face relatively cheap car insurance rates.
[***] Though I gather that there is occasional civil disobedience.
Labels: Economics, externalities, Trains Planes and Automobiles
