Pigou Club Should Favor a 21-Cent Gas Tax, Not One Dollar

In a recent post, we suggested some reasons to be skeptical of Harvard professor Gregory Mankiw’s “Pigou Club” proposal for a $1 increase in the federal gas tax. We argued that those who advocate for taxes aimed at fine-tuning the efficiency of the economy—so-called “Pigouvian” taxes on externalities—had not adequately answered the question of how high such an ideal tax should be.

In a comment on his blog, Mankiw indirectly responded to this criticism. He points to a 2002 paper from Ian W.H. Parry and Kenneth A. Small suggesting that the “optimal” gas tax in the United States is $1.01 per gallon. Currently the combined federal and state average is 45.9 cents per gallon. Mankiw suggests the paper supports his choice of $1 for the optimal Pigouvian gas tax.

Unfortunately, the paper doesn’t justify a $1 Pigouvian tax at all. It only justifies one of about 21 cents—less than half the current federal and state gas tax, and about equal to the current federal gas tax of 18.4 cents per gallon.

Here’s why. The $1.01 figure from Parry and Small isn’t just to correct for Pigouvian negative externalities. It consists of several components, not all of which are Pigouvian pollution taxes. Built into the $1.01 figure are the following:

1. Fuel-related pollution externalities;
2. Distance-related pollution externalities;
3. Costs of traffic congestion;
4. Costs of auto accidents; and
5. An amount to boost gas taxes to equal the optimal “Ramsey Rule” tax.

Only the first two items on this list—pollution from gas per gallon and per mile—are Pigouvian externalities. None of the other items can be properly labeled as negative externalities from gasoline.

Number 3—the costs of traffic congestion—is not a negative externality from gas because the proximate cause of congestion is lack of a system of congestion road pricing to allocate road space, not gas use. Congestion would occur even if cars were powered by solar energy, so it makes no economic sense to impose a Pigouvian gas tax to “correct” for a phenomenon that’s caused by poorly-designed property rights for road space.

Similarly, number 4—the cost of accidents—is not a negative externality from gas for two reasons. First, accidents are fundamentally different from air pollution because they’re fully compensated by collision auto insurance, which is mandatory in most states. Second, like congestion, accidents are not caused by gas use but by driving, and would occur even with environmentally clean fuels.

Number 5 is clearly not a Pigouvian externality tax. That figure comes from the theoretical field of “optimal commodity taxation,” also known as “Ramsey” taxation. Ramsey taxation argues that things with few substitutes like gas should be taxed heavily, while things with many substitutes like labor and work effort should be taxed lightly. Whether or not Ramsey taxation is good policy, one thing is for sure: component number 5 is clearly not a Pigouvian tax, and it’s an error for members of the Pigou Club to count it as one.

The table below shows the breakdown of Parry and Small’s ideal tax for each of these five components. The highlighted lines are the only true Pigouvian taxes (from the full paper at www.rff.org/Documents/RFF-DP-02-12.pdf).

Once the non-Pigouvian components are subtracted, we’re left with an optimal gas tax of only 21 cents—surprisingly close to the current federal gas tax of 18.4 cents per gallon (although that 18.4 cent tax is not a “pollution tax,” but is designed to fund federal highway spending consistent with the so-called “benefit principle“).

By this analysis, Mankiw’s $1 per gallon gas tax would overstate the pollution externalities from gasoline by five-fold. So the question of “how high?” that advocates of Pigouvian gas taxes routinely ignore is far from settled in practice.