Marginal Utility

by Tom Bozzo

A "Dust to Dust" study of the life-cycle energy cost of cars by CNW Marketing Research has made a couple rounds of the intertubes, most recently following an update published this past spring. I caught wind of it by way of a commentary which ran in the Philadelphia Inquirer in April, authored by one James Martin of the 60 Plus Association [*], h/t to Autospies for the link,. Martin tells a horror story (debunked here) regarding pollution near the Sudbury Ontario smelter that supplies Toyota with the nickel that is eventually made into hybrid car batteries, then cites CNW's research for a claim that the life-cycle energy cost of a Prius is $3.25/mile, whereas the energy usage for a Hummer H2 H3 (see update, below) costs a mere $1.95/mile. CNW's claims were also crowed by a Reasonite and parroted by the handsome but not very bright George Will.

Much of the negative reaction focused on the assumption that the Prius would be driven roughly 100,000 miles over its life, whereas the Hummer is assumed to have a 300,000-mile life. The latter figure, I'm sure, is far into the outer tail of the actual distribution of Hummer lifetimes, but that actually isn't CNW's biggest problem. The numerator of the energy cost-per-mile is off by a good order of magnitude, and almost nobody figured it out — not least CNW, who provide sufficient backup material to unearth the order-of-magnitude error and don't seem to realize it.

Economists should be instantly suspicious of a claim that the "cost of energy" for 109,000 miles of Prius driving (including the cost of recycling the Prius at the end) would be on the order of a third of a million dollars. (An Oregonian letter-writer and Mark of Denialism.com, among others, figured out that something was terribly wrong.) A ceiling for the energy's (expected) private cost — in fact, a cathedral ceiling in the MBR of a nice Toll Bros. McMansion — would usually be the price paid for the car plus the bills for fuel, maintenance, and insurance. This is because the various suppliers should pass their input costs on to the end consumer.

In the ballpark of CNW's assumed $3 cost per gallon of gasoline for the actual driving, that would work out to something like of $40,000-$50,000 over the 109,000 miles for a Nicely Equipped Prius. Much if not most of that reflects non-energy input — the metal in the car, mechanics' labor, etc. In fact, analyses showing that most of the life-cycle energy input is mostly in the driving (useful links here from the Union of Concerned Scientists' hybrid blog) would suggest that the energy cost is no more than around $15,000. So the actual private cost per mile is around 14 cents for the 109,000 miles.

CNW asserts that they are not trying to reflect only the pump price of gasoline, but the total social cost of the energy. Can that reasonably account for the differences? As a good liberal, I believe in the existence of substantial external costs of motoring as much as the next person, but I would suggest the answer is No Frackin' Way.

It's tough to put a price tag on those externalities, but let me throw out a figure of $2 trillion a year, worldwide, to cover national security expenditures, pollution controls, CO2 emissions offsets, etc., indirectly related to motoring. Clearly, this can pay for the portion of the worldwide military forces dedicated to securing oil supplies, the Bushist clusterf**k in the Middle East, and leave lots left over.

The $2 trillion is not accidental; it's also roughly the annual bill for worldwide oil demand at the present $65-70/bbl market price. [**] Whatever the real figure, note that there must be enormous external costs to escalate the 14 cent/mile private cost figure by a factor of two, let alone the factor of 20 CNW's Prius cost requires.

The error on the Hummer side of the equation is not as large, though still substantial. The Hummer's energy cost per mile is probably around 2-3 times the Prius's under a more usual accounting. The Hummer's much greater bulk is probably a bigger driver of its embodied costs than the Prius's more advanced materials mix, and on the road it consumes 3-4X the fuel per mile. So if the social cost of the Prius is 30 cents/mile, the Hummer works out in the range of 60 cents to $1/mile.

As I said, CNW could and should have found their error. An example of the information they had at hand is on p. 303 of the report, quoting a Google Answers document that cites a Ford study to the effect that a Taurus-type family sedan would have lifetime energy usage of 961 GJ. A parenthetical note added by CNW claims that the Taurus replacements (Fusion, Five Hundred) would have 1.53 times the energy cost thanks to being "significantly more complex" [***]. That factor is unsourced, but let's grant it: the Five Hundred's energy usage is 1470.33 GJ. How much energy is that, you might ask? The EIA says a gallon of gas has energy content of about 0.124 GJ (it's variable with the fuel blend). Thus, by advanced math, the Five Hundred-type vehicle uses the equivalent of 11,763 gallons of gas. At CNW's assumed $3/gallon cost, that would cost $35,288. Yet for the Mercury Montego (the Five Hundred's Mercury-brand twin), the cost-per-mile and assumed life work out to an energy cost of $344,128. This implies a cost of $29 per equivalent gallon of gas.

Inability to catch order of magnitude discrepancies in your own report does not inspire confidence. CNW's "Dust to Dust" results, in fact, appear to be nearly total bullsh*t.

Update 9/16/07: A rude anonymous commenter points out that the $1.95/mile figure applies to the Hummer H3, not the H2. This doesn't make any appreciable difference to the post, since the H3's fuel economy is 15 MPG city/20 highway by the 2007 EPA methodology, so even the relatively fuel-efficient H3 consumes 2-3x the energy per mile vs the Prius, and its 4,700-lb. curb weight implies higher embodied energy. It was disingenuous for James Martin to pluck the H3 figure out and call it the "Hummer" without a model qualifier; the H3 isn't exactly the "bete noire" of environmentalists in the sense of the CAFE-skirting H1 and H2. For the record, CNW's measured per-mile cost for the H1 was $3.51 and for the H2 was $3.03.


[*] A "nonpartisan" association devoted to the interests of the residuum of senior citizens who have not been made better off by the New Deal and Great Society retirement safety nets.

[**] Which is around 86 million barrels/day.

[***] From the excessive two-digit precision to the potentially mistaken assumption that greater complexity implies more embodied energy, there are lots of uncomplimentary things that could be said about this had I the time.

Labels: Economics, Energy, innumeracy, Trains Planes and Automobiles