Optimal Gasoline Taxes, Given Externalities

A good article on optimal gas taxes is "Does Britain or the United States Have the Right Gasoline Tax?" Ian W. H. Parry and Kenneth A. Small, The American Economic Review, Vol. 95, No. 4 (Sep., 2005), pp. 1276-1289. In 2000, taxes were $2.80/gallon in the UK and $.40/gallon in the USA. They should have been $1.34 and $1.01, in light of congestion, accidents, and Ramsey taxation (with minor contributions from pollution and CO2).

Wikipedia says taxes are $5.20/gallon in the UK, $.47/gallon in the US, $7.61 in Germany, It is important to include value-added tax, which is done in those figures.

The Inst. for Fiscal Studies, more reliable, gives the fuel duty plus VAT per liter in pence for different European countries as from 55 in the UK (the highest) to 24 in Greece (the lowest). Germany is 40; France is 46 (second highest); Italy is 42; Spain is 28.

Thus, it seems Greece and Spain are about at the optimum and all the other European countries are too high.

Curiously, Parry and Small do not mention one of the major arguments for a fuel tax: paying for road construction and repair. I seem to remember that the effect of cars on road deterioration is trivial (it's all due to trucks), but I might be wrong on that, and it seems as if it has to be wrong for city streets.

Parry and Small point out that a gas tax is poorly designed for controlling congestion and accidents, since it is lower for fuel- efficient cars. Also, as implemented, it is invariant across locations, which vary tremendously in the cost from congestion, accidents, and pollution. They calculate the optimal per-mile tax, which does better. That is hard to enforce, though, since if the tax became high, odometer fraud would become common. (Maybe it could be based on how many miles you live from work, though, and age and sex, as insurance rates are.)

What might work better would be to increase the vehicle registation tax, or to at least base the per-mile tax on where the vehicle is registered. Or, we might combine a gasoline tax with a registration fee based on the vehicle's fuel-efficiency, fuel-efficient cars paying a bigger registration fee since they pay a lower gasoline tax per mile travelled.

In practice, I think, hybrids and suchlike are actually subsidized by the government rather than taxed more heavily. What Parry and Small show is that that hybrids would be driven too much, given that they cause accidents just as much as other cars.

In view of the importance of accidents as an externality, I'd like to see that explored more (maybe it is in the paper; I didn't read carefully). A big car is safer for the occupants, but more dangerous to other cars. So it seems, since the effect on other cars is the externality, that big cars should be taxed more.

Parry and Small find the optimal global warming tax to be very small, even using liberal estimates of the effect on global warming and the cost of it. It seems that European countries are emitting too little CO2 from cars, not too much. That result should be publicized. The conclusion is true even if other countries such as the US and China are emitting too much CO2, I think. The cost estimates of the Stern Report and others are based on "business as usual", which means that the marginal benefit to the world from the UK from increasing or reducing emissions is based on other countries' not changing their current policies. Thus, from the point of view of treating all countries neutrally rather than favoring some at the expense of others, the UK ought to emit more carbon dioxide.