The Economist vs. Cato on Gasoline Taxes

With the Pigou Club now having gone silent in the debate over the federal gasoline tax launched by our study of the same last month, the only party left standing to hold the flag for the federal gasoline tax is … The Economist. I hope the former reemerges for a more sustained conversation sometime down the road, but until they do, let’s consider the latest riposte from our favorite news weekly.

The Economist seems to agree that a federal gasoline tax is a very clumsy and inefficient way to deal with the negative externalities associated with driving. So far, so good. But then they claim:

The discussion over taxation of carbon emissions from tailpipes … comes down to arguments over cost efficiency (a gas tax may well be preferable if the cost of assessing emissions for tax purposes is too high).

Maybe I’m not reading this right, but is The Economist seriously entertaining the argument that it might be more efficient to tax carbon emissions via a gasoline tax than via a cross-sector carbon tax? Econometric investigations suggest that it’s twice as costly to reduce carbon emissions via taxes confined to the transportation and electricity sectors than via taxes imposed across the economy as a whole.

Perhaps The Economist meant to say that “The discussion of taxation of conventional pollutants from tailpipes comes down to argument over cost efficiency.” If so, that’s different … and fine. If the costs associated with monitoring and collecting information regarding tailpipe emissions were high, there would be room for an argument for gasoline taxes as a “second-best” remedy (a point we explicitly concede in our paper).

Even so, repairing to a gasoline tax as a second-best means of addressing environmental externalities still does not suggest the need for federal taxation. State or local taxation would be far better for reasons discussed in our paper.

The Economist then moves on to take issue with our most unconventional argument – that even perfect internalization of the negative externalities associated with driving via some set of Pigouvian taxes would make the economy less efficient because it would induce greater use of mass transit. Additional mass transit use is less efficient at the margin because the literature suggests that the discrepancy between costs and charges to users associated with mass transit use are even greater than those associated with automobile use.

The Economist makes four arguments. Argument #1:

The first and most obvious point is that a carbon tax and congestion pricing will quite obviously have the same effect on transit use, other things equal, as an increased gasoline tax … If a reduction in transit use is Mr Taylor's guiding principle, then I fear his prescriptions are misguided.

A reduction in transit use is not my guiding principle. Nor do I prescribe carbon taxes or other emission taxes. We are instead making an If/then argument. IF one wants to internalize the negative externalities associated with driving (pollution, congestion, carbon dioxide emissions, whatever), THEN one should tax those externalities directly rather than indirectly via a tax on fuel consumption.

Sure, we could have stopped there, but many economists embrace Pigouvian taxes because they think they would improve economic efficiency. But according to Cliff Winston’s work, if all transportation options were priced optimally, there would be more automobile use and less mass transit use. Thus perfect Pigouvian taxes on automobile use but no reforms in mass transit pricing (i.e. reforms that would make users pay more of the costs) would have the paradoxical effect of making the economy less, not more, efficient.

So we are giving serious economists – including those of the Pigouvian variety – two very good reasons to oppose a federal gasoline tax. Either one will suffice.

On to argument #2:

Those cited authors [and by this, The Economist is referring to Mark Delucchi and Clifford Winston, whose work we highlight in our study] do make a very compelling case that transit is inefficient relative to automobile use, so long as you take as given the billions of dollars in annual government highway spending and eliminate that spending from your calculations. And that is exactly what the cited authors have done. The Washington Post reported this week that annual federal spending on highways is now over $40 billion, which is, of course, in addition to the billions spent by state and local governments on the construction, maintenance, and policing of roads. Include those figures in the analysis, and the inefficiency argument loses all coherence.

How so? Winston and Delucchi ask the following question; if everything were priced optimally, what would would the cost of driving be? What would the cost of mass transit be? And given those costs, how much of each would people use given the benefits they receive? Both authors conclude that, in a perfectly priced world, there would be less mass transit use than at present and more driving. Hence, if we want to “get the prices right” – the explicit goal of Pigouvian taxation – the result would be more driving.

Now, how does the fact that government has spent billions of dollars on roads affect this exercise? The Economist implies that this was some kind of gift to automobile drivers that Winston & Delucchi are happy to give them without cost. But that’s not true. Until 1975, road construction and maintenance was funded entirely by motorists via the gasoline tax (see Jose Gomez-Ibanez's 1985 essay in a book reviewed here). During the high inflation period of the late 1970s and early 1980s, however, fees and taxes did not keep up with highway expenses. But over the entire 1975-1993 period, the federal gasoline tax increased from 4 to 18.4 cents per gallon (3.6 fold increase) while the GDP deflator increased only 1.3 fold.

The federal gasoline tax has not been increased since 1993, however, and according to our colleague Randal O’Toole, 2003 data suggest that motorists only paid approximately 89% of the total capital and maintenance costs of the highway system during that year.

So there are subsidies to motorists now, but they are tiny on a per-passenger-mile or vehicle-miles-traveled basis. And during the formative years of the highway system, motorists paid more than its costs. Hence, correcting the current imbalance between fuel tax revenue and road construction and maintenance - as Winston and Delucchi do - is sufficient.

Now, argument #3:

That omission is the most egregious, but it certainly isn't the only one. The sources take the structure of the landscape to be exogenous, when in fact highway spending in outer suburbs reduces the cost of commuting from great distances, encouraging outward migration of urban populations. That, of course, then overwhelms existing infrastructure and generates new highway construction still farther out, continuing the outward spread of the city's borders.

We have no complaint with the observation that long distance commuters are not paying the marginal costs associated with their commutes - and neither does Winston or Delucchi. We support getting those prices right via congestion pricing and user fees. and those costs are built into Winston and Delucchi's calculations. But if we’re going to price road use efficiently, we should also price rail use efficiently. And doing so, as we said, would lead to more, not less, driving.

Finally, we arrive at argument #4:

In calculating the positive externalities generated by rail, Brookings Scholar Clifford Winston only considers the effect rail has on reducing highway congestion. But that's far from the only positive externality produced by rail transit. Rail facilitates density in a way that highways cannot, producing and supporting the dense urban networks present in central cities. There is a long and rich literature on the positive externalities produced by such urban agglomerations; it has been suggested, for instance, that a doubling of urban population density leads to a 6% increase in productivity. Certainly that should be included in any consideration of the relative merits of mass transit.

If there are positive externalities generated by rail use beyond the effect of rail use on road congestion, let’s see some citations! We are unaware of any in the peer-reviewed urban economics literature. And remember, subsidized rail also contributes to urban sprawl because it allows people to work in the urban center and travel further out faster at peak rather than living closer to work.

There may be a case for the elimination of the gasoline tax, provided it is replaced with levies on the negative externalities produced by road transport. That case does not include a shift away from mass transit. If the Cato authors wish to make the point that transit use is bad on efficiency grounds, they need to work much harder.

You can lead a horse to water, but you cannot make him drink.