Different math on Cash for Clunkers

In response to those who say that Cash for Clunkers was way too expensive as a way to reduce carbon emissions, Joe Romm has a different way of running the numbers:

Let’s assume the new cars are driven nearly 20% more over the next 5 years, and that the average price of gasoline over the next five years is $3.50. Then we’re “only” saving 140 million gallons a year or roughly $500 million a year. The $3 billion program “pays for itself” in oil savings in 6 years. And most of that oil savings is money that would have left the country, so it is a (small) secondary stimulus.

Using a rough estimate of 25 pounds of CO2 per gallon of gas (full lifecycle emissions), then we’re saving over 1.5 million metric tons of CO2 per year — and all of the ancillary urban air pollutants from those clunkers — for free.

On one hand, I see his point; on the other hand, using this logic, any cash transfer from the government to the population could be considered "a stimulus," absolving the government of any mandate to be cost-effective with its programs. Since this particular "stimulus" pays out gradually over the course of six years, its relevance to the current recession is basically zero. So in conclusion, this isn't the kind of fiscal responsibility I'd like to see the government displaying given the current budget outlook.