A Free-Market Energy Blog

Larrick and Soll’s “Miles Per Gallon Illusion”

By Robert Murphy -- February 13, 2009

Richard Larrick and Jack Soll have started a nifty website to promote their message that the conventional “miles per gallon (mpg)” metric is actually misleading and counterproductive for climate change and energy policy objectives.  In their words:

MPG tricks people’s perceptions. Replacing a car that gets 14 MPG with a car that gets 17 MPG saves as much gas for a given distance as replacing a car that gets 33 MPG with a car that gets 50 MPG (about 100 gallons per 10,000 miles). MPG obscures the value of removing the most inefficient cars. A 14 to 20 MPG improvement saves twice as much gas as a 33 to 50 MPG improvement.

What to do instead?  Rather than measuring distance per volume of fuel, Larrick and Soll recommend measuring volume of fuel per unit of length:

“Gallons per 100 miles” or “Gallons per 10,000 miles” (GPM) corrects these misperceptions. These GPM values should be provided by consumer sites to supplement information about MPG. And GPM makes clear that policy should be focused on replacing the most inefficient cars.

If the difference between the two approaches is not “clicking” with you, I really encourage you to check out this 4-minute YouTube video they created.  Their main message is that thinking in terms of “miles per gallon” is very misleading, and leads to counterintuitive policies if the goal is to reduce gasoline consumption.

For example, suppose that for the same amount of tax dollars, the government can give people incentives to move from vehicles that get 10 mpg to vehicles that get 15 mpg, OR for the same amount of money, the government can coax people to go from vehicles that get 25 mpg into vehicles that get 100 mpg.  Which is a better way to reduce our nation’s dependence on fossil fuels, save the environment, etc. etc.?

If you assume people drive the same number of miles, then the first policy saves a lot more gas.  E.g. let’s say both sets of people drive 6,000 miles per year.  With the 10 mpg vehicle, they burn through 600 gallons.  Switching to the 15 mpg vehicle means they burn 400 gallons, for a savings of 200 gallons per year, per person.

On the other hand, the person starting out with the 25 mpg vehicle burns 240 gallons per year.  When he upgrades to the super-efficient 100 mpg vehicle, his annual fuel consumption drops to 60 gallons, for a savings of 180 gallons per year.

Thus, it saves more gas to coax people to upgrade from their 10 mpg vehicles into ones that get a still-paltry 15 mpg, than it does to coax people to give up their 25 mpg vehicles and get into Star Trek material that achieves 100 mpg.

Now if I might generalize, the larger point here is that there is a danger in using the government to impose uniform policy objectives on the whole country.  After the regulations are in place, it is very hard to make fundamental changes when bright guys like Larrick and Soll notice an elementary flaw with the original design.

In contrast, if private-sector conservation groups had been using their donors’ money to try to “retire” gas guzzlers from the road, it would be very easy for them to refocus their efforts after seeing research like Larrick and Soll’s.  Even if the original non-profits were locked into their ways, it would be straightforward for another group to rise to prominence, and attract donor money once the public learned of the better way to achieve their goals.  This type of adaptation and mid-course correction is much harder when people try to work through the political system.


  1. Fred  

    Cost per gallon vs. cost per mile is also clickable.

    While we buy gas by the gallon we use it by the mile. Cost per mile is a better metric than cost per gallon. For me, the mileage I get using premium gas without ethanol is enough better than ethanol contaminated regular that it is cheaper to use premium.


  2. Ed Reid  

    Regrettably, even “gpm” becomes confusing when plug hybrid vehicles enter the market, since not all of the purchased energy consumed by the vehicle is gasoline, ethanol-contaminated gasoline [ 🙂 ] or diesel. Also, in mock deference to the AGW alarmists, not all of the CO2 emissions resulting from the vehicle’s operation are accounted for in the measure.

    Absent some major technological breakthrough, achieving 100 mpg (or 0.01 gpm) would require plug hybrid technology; and, also, ignoring the energy delivered to the battery from the electric power source, the primary energy efficiency of the power source and the emissions resulting from the generation of the electric power.

    Arguably, a more appropriate measure of the vehicle’s performance would be gallons (gasoline or diesel) equivalent per (vehicle) mile, or “gepm” . From a global energy perspective, the more appropriate measure might be gallons equivalent of primary energy per mile, or “gepem”. From a climate change perspective, however, perhaps the more appropriate measure might be tons of carbon equivalent per mile, or “tcepm”.

    Certainly, if it makes sense to change to a more useful and meaningful measure to replace a measure which has been in common use for decades, it makes sense to change to a new measure which could remain in common use for decades. 🙂


  3. Francis MASSEN  

    To us Europeans being used to l/100km (litres per 100 km) since eons this discussion seems to be of another age 🙂


  4. Ed Reid  

    …but not quite as bad as furlongs per fortnight. 🙂


  5. Tom Tanton  

    it really doesn’t matter WHAT metric is used by the faux-cognicenti. The numbers were not/are not meant to actually represent the mileage obtained (hence the “your mileage may vary” caveats) but ONLY to represent relative mileage (i.e. a car with rating of 20 goes further than one with rating of 19–but will never ever get exactly 20 except instantaneously.) Explaing that to an innumerate population resulted in the current overly simplified and eminently inaccruate syste. It’s not much better than color coding terrorist threat levels.


  6. Will Fuel Economy Mandates Increase Car Company Profits? | Institute for Energy Research  

    […] spell out the argument more fully in this post, but we can see the gist of it with a simple numerical example: Suppose for the same amount of tax […]


Leave a Reply