Let's say you drive 1 mile a year, in a car that gets one mile to the gallon.
* (See
http://en.wikipedia.org/wiki/Reductio_ad_absurdum">Reductio ad absurdum.)
Is it worth it to scrap that old car, and spend all of the energy it will take to manufacture a new car, and ship it to you, so you can drive that one mile a year with the highest efficiency possible? In this extreme case, I think you will agree, it is not.
Now, let's assume you drive 1,000 miles a day, in a car that gets one mile to the gallon. In this case,
I will agree that it makes sense to buy a nice new Prius, or some other car to improve your gas mileage.
OK, so, at some point between these two extremes, there is a break-even point. The difficulty lies in determining exactly where that is. There are multiple variables to be considered. (How bad
is the mileage? How much do you drive? Will you sell the car to another driver or to the junkyard? …)
… He won't commit to exact numbers, but he does have a rule of thumb: If your jalopy is moderately efficient (i.e., gets better than 25 mpg) and you don't drive it much, keeping it is better than buying a new car. "But if you have an old car with pretty lousy fuel economy," he says, "then you're better off getting a new car because the emissions from making a new car are really not that big compared to the emissions from using the car."
…
* (Perhaps it's not that absurd.) When I was a kid, my Grandfather had a Corvair he kept in his barn. When we came to visit, Grandfather would have some errands he wanted to run, but he was too old to drive. So—at Grandfather's insistence—Dad would drive him (a few miles) "in to town" in the Corvair. As a result, that Corvair was only driven a few hundred miles each year.
Though there was no such thing as a Prius in those days, it would have made little ecological sense for Grandfather to have traded in his old Corvair for one.