In an earlier version of this post, the comments thread degenerated into an overheated argument between me and Charles Komanoff, who’s a noted expert in carbon taxes. Mea culpa. I’ve rewritten the post slightly to improve the clarity and, I hope, forestall the antagonism, if not the disagreement

This is surprising: to date, state gas taxes appear to have had very little effect on either driving habits or fuel consumption. Or, more precisely, there’s been no correlation between a state’s gasoline tax and the amount of fuel its residents use or the amount of driving they do.

Don’t believe me? Then feast your eyes on these babies:
gas tax fuel


gas tax vmt

Those are big, fat, and completely uncorrelated blobs. What you’re seeing is all 50 states plus DC plotted to show a relationship between state gas tax rates and per capita fuel consumption (in the first chart) and per capita miles driven (in the second chart). You can see that there is essentially no relationship whatsoever.

Maybe this shouldn’t be surprising. We know that raising the price of gas reduces its consumption. That’s something we all saw clearly during the 2008 price run-up. But there are at least two very good reasons why state gas taxes don’t appear to reduce consumption. For one, the tax revenues are dedicated to boosting consumption. In fact, in most states, gas taxes are set aside especially for building and maintaining roads.

So state gas taxes are sort of like tobacco taxes… if the tobacco revenue were funneled into advertising cigarettes. The tax slightly reduces consumption, but the use of revenue slightly increases consumption.

And for another thing, state gas taxes are pretty small.

  • They’re small enough, anyway, that they are scarcely visible in the price at the pump. Even though there’s a pretty large difference between state tax rates—from 7.5 cents per gallon in Georgia to 36 cents per gallon in Washington — they only account for a small fraction of the consumer price. For a typical state, with a tax rate in the neighborhood of 21 cents per gallon, state gas taxes would account for only about 10 percent of the purchase price—and maybe as little as 5 or 6 percent during the summer of 2008. So although price matters, state taxes aren’t high enough to make a noticeable difference.

    In addition to the two factors I mentioned above — the use of revenue and the small size of the tax rates—there may be dozens of other factors that work to obscure the relationship between tax rates and behavior, such as the relative wealth of states, disparate levels of urbanization, differences in physical geography, and so on.

    There are a couple of lessons, I guess. First, a general one: if you want to use taxes to reduce consumption, you probably shouldn’t spend the revenue to promote consumption. (This isn’t a knock against state gas taxes; they weren’t designed to depress gas use, but to fund infrastructure.) So that means that a carbon tax would still work to reduce carbon emissions as long as the proceeds aren’t used to promote carbon emissions.

    The second lesson has a Northwest policy application and it’s a two-parter. States like Oregon that are considering switching from a fuel tax to a mileage tax should consider the following. Number one, the existing state gas tax doesn’t appear to be doing much to reduce fuel consumption. And number two, taxing mileage at a relatively low level may not do much to reduce driving. That’s in part because Oregon’s proposed mileage tax would be much like existing state gas taxes: it would be designed to support more driving.

    But what does this mean for policymakers? And how can we change the way state gas taxes work? Check back here soon to read Benita Beamon‘s upcoming post to find out. Benita is an expert in road pricing. She’s also an associate professor of industrial engineering at the University of Washington and, we’re proud to say, the newest Sightline Fellow. She’s going to bring her brainpower to bear on this subject and some others.

    Notes: Fuel consumption and driving figures come from the US Federal Highway Administration’s most recent data here and here. Per capita rates are calculated using 2007 population data from the US Census Bureau, here.