David R. Henderson  

Is Mankiw's $2.10 Optimal Gas Tax Correct?

PRINT
What does Greece tell us about... What can we expect from Macron...

In Sunday's post, I noted an important deficiency in Greg Mankiw's treatment of negative externalities and stated that this matters for his treatment of gasoline taxes. Here's why.

Greg lists 3 negative externalities associated with gasoline, and says that the optimal corrective tax is $2.10 per gallon. Here are the three, along with his reasoning for using a gasoline tax for each of the three:

1. Congestion: "A gasoline tax keeps congestion down by encouraging people to take public transportation, carpool more often, and live closer to work."
2. Accidents: "According to the National Highway Traffic Safety Administration, a person driving a typical car is five times as likely to die if hit by a sport-utility vehicle than if hit by another car. The gas tax is an indirect way of making people pay when their large, gas-guzzling vehicles impose risk on others, which in turn makes them take this risk into account when choosing what vehicle to purchase."
3. Pollution. "The burning of fossil fuels such as gasoline is widely believed to be the cause of global warming. Experts disagree about how dangerous this threat is, but there is no doubt that the gas tax reduces the threat by reducing the use of gasoline."

Let's look at them one by one.

1. Congestion. Notice that this is consistent with Mankiw's argument for taxing aluminum because the production process for aluminum pollutes. Just as pollution, not aluminum, is the problem, so congestion, not gasoline is the problem. I went to the 2007 Journal of Economic Literature article that Greg cited. It's by Ian W. H. Parry, Margaret Walls, and Winston Harrington.
Sure enough, they backed my point on congestion. They wrote:

Clearly a fuel tax, which raises driving costs for all regions at all times of day, is a very blunt instrument for alleviating traffic congestion, which is highly specific to rush hour periods in urban areas; the ideal instrument is a road-specific congestion toll that varies with time of day.

They later pointed out: "higher fuel taxes will have a disproportionately large effect on roads with minimal congestion and a disproportionately small impact on congested roads."
It's true that they still advocated higher gas taxes to address congestion, but it would be more accurate to say that they "settled" for gas taxes because of current difficulties in getting congestion pricing, difficulties, by the way, that are diminishing with the advance of technology.

2. Accidents. Think about the careful long-distance driver of an SUV who pays a lot of the tax but causes way less than his pro-rata share of accidents. Wouldn't a better solution be to make those who cause accidents more liable, if they are not sufficiently liable now?

3. Pollution. Greg here is not actually discussing pollution in the normal sense. He focuses entirely on global warming. By the way, notice that Greg goes far beyond even most climate scientists who believe global warming is a big problem. He writes that "the burning of fossil fuels such as gasoline is widely believed to be the cause of global warming." He's wrong. Few people believe that fossil fuels are the cause of global warming. Rather, they are believed to be one of the major causes of global warming.
Back to the economics. The reasoning about global warming, even if it's as bad as Greg fears, justifies a carbon tax, not a tax specifically directed at gasoline.

Interestingly, of the six categories of external costs that Parry et al list, greenhouse warming would lead to a 6-cent per gallon gasoline tax. See their Table 2.


Comments and Sharing






COMMENTS (28 to date)
Dylan writes:

"The burning of fossil fuels such as gasoline is widely believed to be the cause of global warming. Experts disagree about how dangerous this threat is, but there is no doubt that the gas tax reduces the threat by reducing the use of gasoline."

Doesn't this quote say pretty much exactly what you say it doesn't? He talks about fossil fuels, of which gasoline is just one that he is focusing on.

Overall, I think a gas tax isn't perfect for any of the externalities he lists, but I think it is usually one of the better available options. Congestion pricing has major privacy concerns in application, so that seems out. Large vehicles beyond the damage they do in an accident, also do disproportionate damage to the roads per mile driven, and while weight and fuel economy are not perfectly aligned, the correlation is pretty high. And lastly, a gas tax that is aimed at reducing global warming is just one form of a carbon tax. You should definitely also tax the other forms of carbon pollution as well so that there isn't some specific incentive to substitute coal power for gas, but that doesn't mean that the carbon tax on gas isn't a gas tax.

ChacoKevy writes:

2) May I ask you to elaborate #2?

Wouldn't a better solution be to make those who cause accidents more liable
Is the idea then that an understood and upfront greater penalty in case of an accident in an SUV would be more efficient at deterring SUV purchase/usage than a gas tax?

John Hayes writes:

ChacoKevy, if you assume the minimum insurance is reasonable compensation for injuries and the insurance market is competitive then insurance will charge higher rates at a price set by the market eliminating the externality.

This is a better solution because it's durable in the face of change. Even if today there's a correlation between fuel consumption, weight and liability that won't be the case indefinitely.

Once someone creates an electric SUV the externality returns. A Tesla Model X weighs 5200 lbs, a GMC Suburban weighs 5500 lbs and a Toyota Corolla weighs 2800 lbs. The fuel tax is now broken, unless you think it's a good idea for middle income gasoline SUV drivers to subsidize affluent electric SUV drivers.

Taxes on externalities need to think in terms of equilibriums not the disequilibrium at the introduction moment of the tax. The new equilibrium may include changing behavior, or sustaining behavior with a different product.

David R. Henderson writes:

@Dylan,
Good catch. Correction made.
@ChacoKevy,
May I ask you to elaborate [on] #2?
Yes.
Is the idea then that an understood and upfront greater penalty in case of an accident in an SUV would be more efficient at deterring SUV purchase/usage than a gas tax?
No. The idea is that the wrong thing to deter is SUV purchase or usage. The right thing to deter is accidents.

Roger McKinney writes:

What about the positive externalities of driving? I'm really tired of mainstream economists who think only negative externalities exist. They should add the positive to the negative and report the net.

And what about the negative externalities of a gas tax? With a fixed stock of money, people would have to cut back on spending in other areas to pay for the tax.

David R. Henderson writes:

@Roger McKinney,
What about the positive externalities of driving?
I’m open to there being some. I don’t know what they are though. Which ones do you have in mind?

Kevin Dick writes:

@David. I have a technical question about Pigouvian taxes as applied to gasoline.

Assume, in arguendo, that we agree with Parry et al that the fuel related external costs for gasoline are a total of 18 cents/gallon and we limit our analysis to only fuel-related costs (i.e., we assume a separate mileage tax).

Now, we _already_ have specifically directed excise taxes on gasoline. At the Federal level, it's 18.4 cents/gallon. In California, it's 27.8 cents/gallon, for a total of 46.2 cents/gallon.

So we have a specific tax on gasoline that's higher than the fuel related external cost. My understanding of the logic of Pigouvian taxes is that this implies our gasoline taxes are currently too _high_ if we want to achieve the socially optimal consumption of gasoline (again, limiting our analysis to fuel-related external costs).

Do I have this right?

David R. Henderson writes:

@Kevin Dick,
Really good point. I should have noticed it. Yes, I do think you have it right. By their reasoning, the gas tax should be reduced!

James writes:

I have no idea how Mankiw came up with $2.10 but I know he did it wrong unless he took into account some estimate of consumer surplus with and without the tax.

For example, say I make $100 per day and spend all of it on a daily consumption bundle (gas included) for which my willingness to pay is $200. Suppose after the tax goes into effect, my total taxes paid to the government increase by $40 which the government spends on something that I value at $10. I reduce my personal expenditures to a smaller bundle that I value at $125 dollars. These numbers are just an example to illustrate how to set up the problem.

Now in order to be welfare improving, the reduction in negative externalities from my reduced driving would have to be greater than the reduction in consumer surplus that I experience after I adjust to the tax.

Since Mankiw has no way to know how much consumer surplus anyone would lose from having to pay the gas tax, or how much anyone really values the way the government would spend the money, I do not see how he can claim to know what level of gas tax is optimal.

AS writes:

James that is a good argument and one often overlooked. Any new tax is likey to just get squandered on more wasteful government spending. Only a revenue-neutral or better revenue-decreasing tax should be considered. Far better to tax gasoline than income.

Kevin Dick writes:

@David. Cool. I've always thought so and it's nice to know that a practicing economist agrees.

Another point that I try to explain in these debates is that it's very important to get these social cost estimates as accurate as possible.

Some people seem to believe that it's desirable to "err on the safe side" of a larger social cost estimate. Obviously, as a theoretical matter, this is wrong for economic efficiency.

Moreover, we're actually talking about a practical cost in _lives_. The desperately poor are especially vulnerable to energy prices. At the margin, higher cost energy results in statistical deaths and fewer QALYs in such populations. I once attempted to find literature quantifying the relationship, but was unsuccessful. If you run across any, please post it.

It would be nice to overcome the psychological bias towards "doing something" with a concrete estimate of people that could be killed by overly aggressive attempts to reduce fuel consumption. Kind of like pointing out that the CAFE standards statistically kill people by forcing manufacturers to build lighter weight vehicles.

mariorossi writes:
No. The idea is that the wrong thing to deter is SUV purchase or usage. The right thing to deter is accidents.
Isn't this a bit inconsistent with your arguement we should tax pollution and not production? Number of accidents is not really our target either: total damage from accidents is. If there is a positive correlation between damage and type of vehicle involved in an accident, taxation that pushes towards less risky vehicles is more in line with your point I would say.

To be honest, I am a bit confused on why this is an externality though. If you are liable for the damage you cause in an accident and that liability is the base for your insurance premium, why is it an externality?

I guess you could argue that it makes other people mistakes more expensive for them as well. If I cause an accident (e.g. cross the road without looking) and the other driver is in a SUV, I am going to end up a lot worse. Is that the argument?

Cullen writes:

I'm going to presume that I am much more likely to die if hit by an 18-wheeler than by a one seater smart car. Or even a compact car than a tiny smart car. My question then is not whether bigger vehicles are more lethal when in accidents (that seems obvious) but if bigger vehicles are in more lethal/serious accidents than small cars are.

David R. Henderson writes:

@James,
I have no idea how Mankiw came up with $2.10
He got it from the JEL article I referenced and linked to. See their Table 2 at the end.
but I know he did it wrong unless he took into account some estimate of consumer surplus with and without the tax.
Good point, and well exposited in the rest of your piece. One would like the funds to go to those who are damaged but that almost never seems to be what the Pigouvians advocate.
@Kevin Dick,
Thanks. Well stated. I recommend that everyone who is still paying attention read Kevin’s comment.
@mariorossi,
Isn't this a bit inconsistent with your arguement we should tax pollution and not production?
Yes. Good catch.
Number of accidents is not really our target either: total damage from accidents is.
Exactly. Now apply that same reasoning to the aluminum discussion where the rest of us failed to persuade you.

David R. Henderson writes:

@mariorossi,
To be honest, I am a bit confused on why this is an externality though. If you are liable for the damage you cause in an accident and that liability is the base for your insurance premium, why is it an externality?
Again, a good point. That’s why I wrote:
"Wouldn't a better solution be to make those who cause accidents more liable, if they are not sufficiently liable now?”
Note that I left open the possibility that they are sufficiently liable.

mariorossi writes:

I still think I am not able to explain my point.

You stated that fixed proportions are required to find a solution. I still mantain that any monotonic relationship would guarantee a solution. Non monotonic relationship could introduce multiple solutions but you would still have them.

Finally I disagree we can just assume more information. Gathering information has costs.

Kevin above proposes a mileage tax. I see the attractivness of it, but it seems very difficult to gather the information. It would require quite a lot of infrastructure/survaillance. I can see how it would be possible in the robot-car future, but it seems still a bit premature.

If the relationship between externality and observable is relatively stable and if measuring the externality directly is difficult accepting the error can still be better than doing nothing.

Rob writes:

justifies a carbon tax

I don't see that. I am not sure a pigovian tax is ever justifiable.

Any time I see something like that, I wonder what Coase's response would be.

MikeP writes:

By their reasoning, the gas tax should be reduced!

This is a point I have been making for some time.

If you want to see the difficulty in getting this concept across to others, you can take a look at this thread from 2009.

Rob writes:

MikeP,

Of course you had problems getting it across, you were arguing with Neu.

Stephen Gradijan writes:

Regarding lost consumer surplus from "Pigouvian taxes":

I don't know what Mankiw proposes in his textbook(s), but on his blog I have seen several times (all times?) whenever he has advocated "Pigouvian taxes" he has mentioned that he wants to have those taxes offset by tax cuts elsewhere, and I believe (but I am not sure) that his preference is for payroll tax cuts.

MikeP writes:

I am not sure a pigovian tax is ever justifiable.

It does take a great leap to accept a Pigouvian tax, as evidenced somewhat in that thread. Somehow the collected tax monies have to actually disappear from all sight -- be laundered, if you will -- in order for it to meet the Pigouvian objective. For if we watch the collected tax we might see it being rebated exactly to those who paid it, or we might see it being used 100% to climate change damage remediation, or we might see it paid out as a dividend to all in order to buy the votes to keep it alive. Surely each of those has a great and very different impact on economy, environment, and policy.

Any time I see something like that, I wonder what Coase's response would be.

My best guess for the eventual resolution of the greenhouse gas problem is that sometime in the next few decades the technology will be developed to allow industrial scale sequestration of atmospheric CO2 at a cost lower than the social cost of carbon. Think great belts coated with nanoparticles that capture CO2 in the light and release it in the dark churning all day long in the desert. Or biotechnology that builds fuels out of atmospheric CO2 using solar power.

At that point the whole problem can be solved by Coase. Each producer of CO2 must pay a reducer of CO2 to remove the equivalent CO2 from the atmosphere that the producer emits. The costs of sequestration will only get lower as technologies improve, and the social cost of carbon in the atmosphere will only get higher as humanity gets wealthier, so this will rapidly become a bargain and a minor expense in what is otherwise business as usual production of energy.

Tionico writes:

Mike P says: Think great belts coated with nanoparticles that capture CO2

the same God who invented carbon dioxide also invented, and abundantly distributed, precisely the "system" you advocate. They are called "plants". It always astounds me how so many alledgedly educated individuals have either never learned (more likely these days, considering our gummit skewlz) or have forgotten (or wilfully ignore) tne simple "carbon cycle" as taught when I was in about sixth grade. The CO2 in the air it taken up by plants, and is THE sole soure of the carbon plants fix into starches, sugars, fats, cellulose, as they grow, all done by solar energy. In return those plants release pure oxygen, returning that into the atmosphere for animals to breathe, which oxygen then metabolises (oxidises, burns) those starches, fats, sugars, in the plants. There is, in the earth today, not one more molecule of carbon than there was when god made this earth. HE designed a perfectly balanced and sustaining system. Your pissant tax schemes cannot change a thing. the "fossil fuels" we burn today are dead plants full of carbon. More plants are dyiung daily, sequestering their carbon into the ground, forming more "fossil fuels" for some later time.

the oceans also play a huge part in maintaining CO2 levels in the atmosphere... and it is the oceans that are the leading factor, not CO2 presence. Learn your science. The Climate shills are the science deniers.

Tionico writes:

The concept of fuel taxes equalising the differences in "harms" done by heavier vehicles is nonsense. States already do this by imposing a tax directly upon the heavier vehicles through registration and "use tax" fees. Further, my insurance company has used their actuarial tables to determine the "risk of damage" factor my vehicle imposes doe to its increaded weight. They use those same tables to calculate their theoretical risk exposure based on my driving history (perfect for the many decades and some two million miles I've driven), mostly cancelling out the theoretical risk of greater damage in case of a crash, which I've NEVER been involved in.

FUrther, my heavy vehicle gets far better fuel economy than others in the same weight class... the fuel tax "unfairly" burdens drivers of similarly heavy vehicles that get poorer mileage.. the same exact body/chassis I have, powered by a different engine, will use up to twice the fuel mine does. Why penalise THEM and let ME skate? (I'll take it.. part of why I INSISTED on the powerplant fitted to mine).

But my REAL beef with turning to higher fuel taxes to "cure" so many of the issues such as maintaining/building highways is this: I drive a lot, and see some form of constructionmaintenance on nearly every trip. As I observe these worksites, and the rules, requirements, etc, for tem, I see SO MUCH WASTE>.... cubic tax money squandered in stupid senseless ways that produce NOTHING except to force the expenditure of far more man hours then needed to get the job done. I am also well aware that ANY publically funded constructioin project mandates "prevailing wage" be paid to all workers, and even requires considerable "make work" for the contractor to verify/prove that such wages were actually paid to all involved. Thus a young kid pushing a broom or picking up scrap/waste MUST be paid a minimum of $36/hour (my state), AND there are many rules imposing specific requirements to "employ" so many to fill unneeded positions. This is one more "benefit" of union jobs, the union promotes rules that result in employing far more workers, AND restricts what any given worker can do... a guy who operates a backhoe CANNOT hop off his machine and use his hands to pick up a rake and use it for a few seconds to do something the machine cannot readily do. No, the operator MUST sit, idling the machine, call for a different class of worker who is "qualified" and "allowed" to go find a rake, saunter back over next the backhoe, scratch the dirt as needed, thenn saunter back out of the way to let the 'hoe operator finish his task. I'd say fully thirty percent of the billed man hours on highway construction projects are "padding", and each one at a minimum of $36/hour before benefits/taxes/costs (closer to $50 hour in total). I'll also mention the mandating of miles of orange plastic fencing installed that does NO benefit, and most often gets left behind polluting the counryside long after the jog is "done", or the stationing of State Police with their cruisers, blue blinking lights flashing away while they read a book or surf the net on the in car laptop, adding about $150 per hour each to the total cost of that job.. last night on a hundred mile trip I noted THREE such trough sloppers.... let's see, $450/hour times a ten hour duty is close to $4500, PER night..... a 30 day project wastes $135 THOUSAND.. not counting overtime/night duty pay.

When THESE types of waste are addressed, THEN we can think about increasing the tax on motor fuel.

Jerry Miller writes:

With respect to item #2, I plan to keep driving my aging, gas-guzzling SUV as a defensive measure, although SUVs are more prone to rollovers. Driving a car when there are so many SUVs on the road is like taking a knife to a gun fight.

S D writes:

Greg should visit France where about 60% of a litre of fuel is tax of some form. Cars are very small by American standards and car ownership per household is far lower than for American households of a similar income level.

Despite the high tax, regular roads in and around Paris are very congested at peak times. By contrast - France's network of tolled motorways - are generally clear.

MikeP writes:

Despite the high tax, regular roads in and around Paris are very congested at peak times. By contrast - France's network of tolled motorways - are generally clear.

From an earlier comment...

...the current gas tax is higher than the social cost of carbon carbon tax would be. And the gas tax in Europe is many multiples of the social cost of carbon.

Are your roads less congested now? Are theirs?

A carbon tax would be between 1 and 2 cents per mile, maybe 3, depending on your car's miles per gallon.

The congestion-priced toll lane on I-680 in the Bay Area doesn't even charge less than 3 cents per mile. It maxes out at 65 cents per mile.

The carbon tax applied to gasoline is simply an order of magnitude lower than the massive consumer benefit of driving.

[In hindsight, I should have said "two or three orders of magnitude lower".]

Mark Bahner writes:
Clearly a fuel tax, which raises driving costs for all regions at all times of day, is a very blunt instrument for alleviating traffic congestion, which is highly specific to rush hour periods in urban areas; the ideal instrument is a road-specific congestion toll that varies with time of day.

In 20-30 years, virtually all travel with by autonomous vehicles. This will strongly promote "transportation as a service," wherein people will not own their cars (and therefore have their cars sit depreciating for 20+ hours every day), but will instead pay for cars the way we currently do for airline tickets.

With autonomous vehicles owned by fleet owners, it will be possible to know essentially where every car is every second of the day. This will allow extremely "intelligent" tolling of every single road.

Roger McKinney writes:

The positive externalities of driving go back to the early days of autos, they replaced horses as the main source of transportation. Growing food for horses took up half the farmland and so raised the prices of food. Also, horses filled up cities with manure which became pulverized and hung in the air for people to breath. Then there was tetanus spread by horses.

Transporting goods became much easier and cheaper with motor vehicles. Less food spoiled. So people got food cheaper as well as an expanded menu of food from greater distances.

Today people can live farther from their work which means less congestion in living spaces and lower housing costs. At the same time, the bedroom communities in which commuters live benefit from having them live there.

POST A COMMENT




Return to top