The MPG illusion: why 15 to 20 saves more fuel than 40 to 50

A household owns two cars. They can upgrade exactly one of them, and both upgrades are available for the same money.

The first car does 15 miles per gallon and could be replaced with one that does 20. The second does 40 and could be replaced with one that does 50 — a bigger improvement on paper, ten extra miles per gallon rather than five, and a much better-looking headline figure.

Both cars are driven 10,000 miles a year. Which replacement saves more fuel?

Almost everyone says the second. The first saves more than three times as much.

Miles per gallon answers the question "how far does this car go on a fixed amount of fuel?" That is a perfectly sensible question. It is just not the question anyone is actually asking when they compare cars.

What you want to know is the opposite: how much fuel does this car need to cover a fixed distance? You do not set out to burn a gallon and see where you end up. You have a commute, and you want to know what it costs.

Those two questions are reciprocals of each other, and reciprocals are not linear. Adding five miles per gallon does something quite different at the bottom of the range than it does at the top, because you are adding to the denominator of a fraction rather than to the number you care about.

The consequence is that MPG compresses the numbers exactly where the savings are largest. The gap between a 15 MPG car and a 20 MPG car looks the same on paper as the gap between 40 and 45 — and in fuel, it is six times larger.

Fuel used over a fixed distance is distance divided by MPG. For 10,000 miles a year:

At 15 MPG: 10,000 ÷ 15 = 667 gallons. At 20 MPG: 10,000 ÷ 20 = 500 gallons. The upgrade saves 167 gallons.

At 40 MPG: 10,000 ÷ 40 = 250 gallons. At 50 MPG: 10,000 ÷ 50 = 200 gallons. The upgrade saves 50 gallons.

Nothing subtle is happening. It is one division, done twice. The illusion is not in the math — it is in the fact that MPG puts the thing you care about underneath the division line, where changes to it stop behaving the way changes normally behave.

The practical consequence runs against instinct. If a household has one thirsty vehicle and one efficient one, the fuel savings available from improving the thirsty one are far larger — even when the efficient one could be improved by a bigger MPG margin.

A pickup going from 15 to 18 MPG saves 111 gallons over 10,000 miles. A hatchback going from 40 to 50 — a much larger improvement by any MPG measure — saves 50. The pickup wins by more than two to one, on a smaller headline change.

The same logic applies to fleets, and this is where the effect stops being a curiosity. An organization replacing its least efficient vehicles first will save considerably more fuel than one chasing the highest MPG figures it can buy, for the same money. Larrick and Soll made precisely this argument when they named the effect in 2008.

Invert it. Measure fuel per distance rather than distance per fuel, and the numbers start behaving themselves: they scale linearly with cost, they can be compared by subtraction, and the savings stop hiding.

Most of the world already does this — liters per 100 kilometers is the standard nearly everywhere outside the United States, and it is the same idea. The US equivalent is gallons per 100 miles, which the EPA now prints on the fuel economy label alongside the MPG figure for exactly this reason.

Once you are working in gallons per 100 miles, comparing two cars is subtraction. A car at 6.67 gallons per 100 miles and one at 5.00 differ by 1.67 gallons for every hundred miles driven, and that difference is directly proportional to what you will pay. No reciprocal, no compression, no illusion.

Every figure above assumes the rated MPG is what the car actually achieves, and it is not. EPA ratings come from a standardized laboratory procedure, and real economy departs from it with speed, load, terrain, weather, tire pressure and traffic. Sustained driving much above 60 mph degrades it noticeably.

That does not undermine the comparison — the reciprocal relationship holds whatever the true figures are, and both cars are measured the same way. It does mean the absolute gallon counts should be read as the shape of the answer rather than a promise about your driving.

Fuel is also only part of what a vehicle costs. Tolls, insurance, maintenance, tires and depreciation are all outside this, and depreciation in particular usually dwarfs fuel over the first few years of ownership. A replacement that saves 167 gallons a year is saving about $583 at $3.50 a gallon, which is a real number and not necessarily the deciding one.