Skip to content

Driving Taxes for the 21st Century

Both Max and Lucas have recently written on this blog about the need to price gasoline appropriately. I agree with them…mostly.  I mean, how could I disagree with them? I’m the one driving the gray Prius with the license plate “TAX GAS”.  But, as I and the others who have advocated for higher gas (and diesel) taxes have recognized all along, it is an imperfect way to price the externalities of driving.  And it is likely to get worse.


A good idea, but not the whole solution

For more than a decade, the students in my MBA course on Energy and Environmental Markets have listed the externalities from driving and then discussed how well taxing gasoline prices those externalities. The list usually looks something like this:

  1. Greenhouse gas emissions
  2. NOx, particulates, and other local pollution emissions
  3. Energy security
  4. Congestion
  5. Accidents

The students generally get the emissions externalities right away, and the energy security externality pretty quickly. Congestion externalities — my decision to get on the freeway slows down all the other cars on the road — sometimes takes a bit longer.  Accident externalities — my decision to drive increases the chances that another car will hit or be hit by me — are almost always the last to be pointed out.  Most students are surprised to learn that congestion and accidents are the largest externalities from driving.


Slowing down other cars is often the largest negative externality from driving (

Then we get to the livelier part of the discussion: is taxing gas an effective way to have drivers internalize the externalities that they create?

Before I discuss the answers, let’s recognize that no public policy perfectly targets the problem it is meant to address. Every tax break is utilized by someone it was not intended to benefit and goes unnoticed by someone whose behavior it would have changed in exactly the hoped-for way.  Subsidizing the purchase of an energy efficient refrigerator sometimes causes a household to go from having one refrigerator to two, keeping the old one running in the garage or basement.  Fuel economy standards get people to buy more efficient cars, but don’t encourage them to drive any less.

Still, even if perfection is unreachable, we need to understand the policy imperfections and work to improve them.

The discussion of gas taxes and greenhouse gas emissions is always very satisfying, because it turns out that the correlation between burning gasoline and emitting GHGs is nearly perfect. Every gallon results in about 20 pounds of CO2 emissions.  So, if you want to put a price on GHGs, taxing gasoline is pretty much the same thing when it comes to emissions from gasoline-powered cars.  One smiley face for gas taxes. 


Much of the problem comes from a small number of old smokies (Oak Ridge National Laboratory)

But the students also start to see red flags as they apply that logic to the other categories. The high correlation with GHG emissions evaporates when it comes to NOx and other local pollutants. These emissions, which contribute to ozone and other health-damaging pollutants, have a very low correlation with the amount of gasoline the car uses. Old cars are massive polluters compared to new cars, due to great technology improvement in pollution control systems. And even within the same year and model, there is huge variation in the quantity of these emissions, as our MIT colleague Chris Knittel has shown in work with Ryan Sandler. Taxing gasoline is not an effective way to go after the small share of cars that put out most of the local pollution.  A frowny face for gas taxes.

Energy security is always a bit hard to explain, but it generally means some combination of greater risk to our economy when we import a lot of oil, and greater security risk when oil sales enrich the autocratic leaders of oil exporting countries.  As US oil production rises and world prices fall, it’s less clear that this is a big externality, but it is clearly still highly correlated with the amount of gasoline one uses.  Another smiley face, though probably a less important one.

By now the students see where this is going, despite the fact that I have told them of my license plate at the beginning. Congestion is also likely to be poorly correlated with the amount of gasoline a car burns. Some people drive on crowded freeways at rush hour, while others drive on uncongested roads or at off-peak times. I’m not aware of any good studies on the variation in congestion externalities across drivers, though someone at Waze/Google should be able to tell us a lot on the subject. That one almost certainly gets a big frowny face.

Accidents are more complicated because some of this externality is internalized through your insurance rates. But work that Max has done with Michael Anderson points out that insurance does a poor job of internalizing the accident-risk externality, because of low insurance requirements and limited cases of liability.  Max and Michael find that a gas tax does a pretty good job of representing the fact that heavier cars are more likely to hurt other people, but it still doesn’t capture the variation in where and when people drive, or much of the variation in how carefully they drive.  Hard to know for sure, but gas taxes probably aren’t great.

Perhaps the most interesting part of this debate is not how well taxing gas captures externalities today, but how that will change in the next decade or two. Gas taxes will almost certainly remain an excellent way to price greenhouse gas emissions and, to the extent they are relevant, energy security externalities.

Technology, however, is increasingly giving us much better ways to address the other externalities, though not without their own issues. Onboard computers will be able to inexpensively monitor tailpipe emissions so we can know exactly how much pollution a car has put out in the last year (though tampering with the equipment may still be a problem – see the VW debacle).  GPS will be able to report to that computer how many of the miles were driven on roads that Google was coloring yellow or red at the time and, with some sophisticated algorithms, even calculate how many delay minutes you imposed on the drivers around you.


GPS could easily tell your onboard computer when you’ve been driving on congested roads (LA freeways at 7:13am today)

At the cost of a modicum (OK, a whole lot) of privacy, we could price pollution and congestion externalities to an extent that perhaps only an economist could love.  The difficult conversations we have been hearing lately about the trade-off between privacy and social responsibility will come to vehicle transportation.

And the possibilities for pricing accident-risk externalities are even more exciting/disturbing.  That onboard computer will know how close you came to hitting the other car or tree or pedestrian, as well as every time you accelerated too quickly or braked too hard.

By now, you may be thinking, “wait, when onboard computers are monitoring that much information, they will also be driving the car.”  Maybe so.  But given the blowback I’ve heard from drivers who seem to think that the right to drive old polluters is also protected by the Second Amendment, I don’t think see everyone giving up their vehicular autonomy any time soon.

And sometime in the next decade we will have to face up to the fact that electric or hydrogen or biofuel powered vehicles have the same congestion and accident effects as the ones powered by hydrocarbons.  As they become a larger share of the fleet, gas taxes will become even less effective for these major externalities, though still a fine way to capture GHG emissions.

In 20 years, if vanity license plates aren’t obsolete, I will have to get a new one: MEASURE AND TAX ALL DRIVING EXTERNALITIES.  Hmmmm. That may have to go on my LED bumper sticker.

ADDENDUM: As commentors have pointed out, wear and tear on roads is also an externality, because we don’t pay for the damage our vehicles do to the roads.  That is clearly correct.  There seems to be some disagreement about how much road damage increases with weight (though it is recognized to be more than proportional), and about how much damage occurs due to weather apart from vehicle use.

Severin Borenstein View All

Severin Borenstein is E.T. Grether Professor of Business Administration and Public Policy at the Haas School of Business and Faculty Director of the Energy Institute at Haas. He has published extensively on the oil and gasoline industries, electricity markets and pricing greenhouse gases. His current research projects include the economics of renewable energy, economic policies for reducing greenhouse gases, and alternative models of retail electricity pricing. In 2012-13, he served on the Emissions Market Assessment Committee that advised the California Air Resources Board on the operation of California’s Cap and Trade market for greenhouse gases. He chaired the California Energy Commission's Petroleum Market Advisory Committee from 2015 until its completion in 2017. Currently, he is a member of the Bay Area Air Quality Management District's Advisory Council and a member of the Board of Governors of the California Independent System Operator.

22 thoughts on “Driving Taxes for the 21st Century Leave a comment

  1. Lists imply comparability:

    1. Greenhouse gas emissions
    2. NOx, particulates, and other local pollution emissions
    3. Energy security
    4. Congestion
    5. Accidents

    “Externality” is a useful concept, but the premise of this piece is that we gain more understanding than we lose by equating, to pick two, GHG emissions and congestion. To put it bluntly, who cares if significant progress on reducing GHG emissions comes with inefficient congestion management? The former is simply much more important.

  2. Once you enforce catalytic converters not just for new but also existing cars (or charge a flat-rate fee for the right to drive without), the local pollution problem is pretty much confined to diesel. So you could place an a local pollution tax on diesel, in addition to the GHG tax. And since most trucks run on diesel, you could also add a “road damage” tax to diesel. So in both cases, you want a higher tax on diesel than on gasoline, to get rid of some frownyness…

  3. In graduate school (in economics at the University of Toronto), I was taught (obviously as a first approximation) that the need for roads (i.e., to reduce congestion, open up new areas) is caused by cars, but road maintenance is largely caused by big trucks. Maintenance requirements increase more than proportionally with weight, so the difference between big and small cars (and small and medium trucks for that matter) is real, but the difference between any car and any 16-wheeler is extremely large. And to add another wrinkle, street cuts to install/repair/replace underground utilities cause a large chunk of the need for urban street maintenance (a study I did for the City and County of San Francisco almost 20 years ago).

  4. Michael Anderson’s comment reminds me of a “highway cost allocation study” I worked on early in my career, bringing the experience of doing electric utility cost allocation studies.

    There are several parts to the truck cost allocation. The first, he notes, is that the roadway damage is an exponential function of axle weight exceeding the elastic limits of the pavement.

    The second is lane width and right-of-way requirements. 102″ transit vehicles and trucks require more lane width than passenger vehicles.

    The third is the policy, nearly universal in the US, of taxing gasoline (105,000 BTU/gallon) at the same cents per gallon as diesel fuel (138,000 BTU/gallon, mostly due to higher carbon content, [The carbon content issue also raises an environmental issue unrelated to roadway wear.] So even if the ton-miles per million BTUs were the same, the $/ton-mile would be lower for diesel vehicles.

    Add to that the 60% higher efficiency of diesel engines at turning a million BTU into propulsion (8 mpg for a diesel semi vs 5 mpg for a gasoline semi), and the $/ton-mile equation is even more skewed, with diesel vehicles paying about half as much roadway tax per ton-mile (in every vehicle classification) as gasoline vehicles.

    Finally, in many states, a portion of the highway revenue comes from per-vehicle taxes, not per-gallon taxes. In my town, the City levies a $20/vehicle/year tax on every motorcycle, car, truck, and heavy truck; the money goes for roadway maintenance. In other places, there is a motor vehicle excise tax, based on value, so a Lexus pays more than an F250. Needless to say, a heavy truck operating 300 days a year doing local freight delivery does a bit more damage than a Prius with a Tax Gas license plate.

    The “electric cars get a free ride” calculation is also complex. In my city, electricity is subject to a 4% state utility gross revenue tax and a 9% city gross revenue tax. Both of these go to the general funds for schools, parks, cops and firefighters. Gasoline is exempt from these taxes, because the state constitution limits gasoline taxes to highway purposes. So, while electric cars may not pay the “road tax” gasoline cars don’t pay the “school tax.”

    Defining “equity” in this context is a job for priests, philosophers, and politicians, not for economists. Our job is to point out the folly in incomplete arguments.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: