(Lots of) Cash for Clunkers

Edmunds.com reports that its statistical analysis of the Cash for Clunkers program finds that the program generated only 125,000 extra new vehicle sales, meaning that the cost to the U.S. government was $24,000 for each of those new cars.

The reason the cost per incremental car is so high is that, according to Edmunds.com’s modeling, 82 percent of the vehicles purchased under the program would have been bought this year anyway, even without the subsidy. So Cash for Clunkers mostly just turned out to be a gift from the government to people who happened to be in the market for a new car at the right time. The auto manufacturers and dealers did not end up getting a very big chunk of the money ultimately, although they did get paid earlier rather than later in the year.

Is this surprising? Not to an economist. It is relatively easy to move around the timing of when someone purchases a durable good, but much harder to affect whether they buy a durable good or not.

For the second time in a week, I am deeply disappointed at the response of the Department of Transportation to research into areas of relevance to the department. The first case was Secretary LaHood‘s response to my research on car seats. Here is what the agency had to say in response to the Edmunds.com analysis:

“It is unfortunate that Edmunds.com has had nothing but negative things to say about a wildly successful program that sold nearly 250,000 cars in its first four days alone,” said Bill Adams, spokesman for the Department of Transportation.

The right response, it seems to me, is either to say 1) that this new evidence convinces us not to do the program again, or 2) that this analysis is wrong. That’s the response that Macon Phillips had on the White House blog (who knew the White House had a blog!):

The Edmunds analysis rests on the assumption that the market for cars that didn’t qualify for Cash for Clunkers was completely unaffected by this program. In other words, all the other cars were being sold on Mars, while the rest of the country was caught up in the excitement of the Cash for Clunkers program. This analysis ignores not only the price impacts that a program like Cash for Clunkers has on the rest of the vehicle market, but the reports from across the country that people were drawn into dealerships by the Cash for Clunkers program and ended up buying cars even though their old car was not eligible for the program.

I’m not sure whether this argument is empirically important or not, but at least it is actually engaging in a meaningful way with the Edmunds.com analysis.


You did not of course factor in the environmental benefits of taking old cars off the road.

Jason S

I just wonder what the reaction will be when everyone that participated in the Cash for Clunkers finds out that they have to pay taxes on their "free money" gift. Many of these people that couldn't have purchased a car without the program may soon find themselves getting it taken away with to pay for their taxes.
Wonder how much that will stimulate the economy?

Gene S

Edmunds analysis is unsatisfactory for two reasons. The first is that, until Cash For Clunkers, car sales were terrible. Their analysis of previous data just ignored the then-current trend. Second, Edmunds ignored the fact that selling is an emotional business. When salesmen are confident, they sell more cars. The C4C program injected confidence into the equation. It also helped that C4C created traffic in showrooms which had become ghost towns.

I expect an analysis of all car sales with and without C4C could be made which would show that the actual cost per resulting added car sale might be as low as $1500, and be no less valid than the Edmunds analysis.


I still want to know how a 1998 Dodge Neon and 2008 Hyundai Accent qualified as trade-ins for the program. These two showed up on the CARS pdf list of traded in vehicles.

The '98 Dodge would never have had less than 18mpg and was never listed on the CARS websites as qualifying The 2008 Hyundai was actually listed on the CARS website as qualifying yet the EPA website had it's mileage at least at 28mpg for the automatic. And that is not even posing the question of why someone would trade in a 2008 vehicle anyhow...

Basically the taxpayers gave at minimum $7,000 of false refunds for those two cars...


You did not of course factor in the environmental benefits of taking old cars off the road.

— Craig

Eric Morris had a good piece on this a while back. How do we know that people with their new fuel-efficient cars won't just drive more?


I agree with you Craig that it should be considered. I'd really love to know the cost per carbon unit reduced. I have a feeling there would be a cheaper way to get the same result.


They should also factor in the environmental COST of the program.


I find this form of "accounting" annoying regardless of the politics. Say I take my two kids out to dinner - I buy them each a meal for 5 bucks each. One isn't hungry. Does that mean I spent $10 bucks "each"? since one meal didn't get eaten?

Christopher Strom

@ craig (#1)

...or the envronmental costs of manufacturing the new cars to replace them.

Nothing's free.


Craig does have a point: while 125K marginal extra cars were sold, Edmunds does state 690,000 clunkers were redeemed. I understand that to get the voucher, the dealers have to proof the destruction of the engine (other parts can be reused or scrapped). If the program yields 690,000 vehicles with, say, 10 MPG better efficiency, is that better or worse than other environmental spends? I will assume 12000 miles per year, from 15 MPG to 25 MPG. This fleet then will drive for 8.2 billion miles per year, consuming 331 million gallons of fuel instead of 552 million gallons, a saving of about 220 million gallons of fuel. As a gallon of gasoline produces 19 pounds of CO2, that's about 2 millon tons less CO2 per year. How's that compare for the price to the feats of geoengineering that got all the climate scientists up in arms with you guys?

Mike B

A better programme would have been to put a question out on the tax return or some such asking if the person intended to buy a car or not in the next 12 months without any explanation. Then when cash for clunkers came around only those who has answered 'no' would have been eligible for the programme.

Of course this only works once as people will quickly learn not to be honest about such things.


Here's some empirical evidence: my sister-in-law, after visiting the dealership and identifying a replacement for her high-mileage SUV, found that her vehicle wasn't eligible for the CfC program. It was too late however; she got the "new car bug," bought a used one at a local dealership, and sold her old car private-party to her friend. So you have two sales motivated by CfC which weren't statistically captured.

Nick J

Only one way to describe the C4C progam:

Frédéric Bastiat's "parable of the broken window."


Here's another impact of the "cash for clunkers" program: it drove down the prices for used cars. I just moved to a metro area and and to sell my car last summer. It lost more than $2000 in value in a couple of months, no doubt due to the diversion of used car buyers into the program.


Is it really better to make a whole new car rather than just repair an old one, though?


Craig -- that would be a much bigger factor if the Cash for Clunkers program actually encouraged people to buy fuel efficient vehicles. I believe that the mpg disparity between what qualified people to buy a new car was too small to make a noticeable environmental impact.


@Craig - And you did not factor in the environmental costs of creating brand new cars to sell.


"You did not of course factor in the environmental benefits of taking old cars off the road."

Not sure of the accuracy, but one analysis I heard indicated that all of the environmental impact could be replicated by turning off American manufacturing for one hour.

You also have to consider the environmental impact of trying to artificially create increased demand...how many years could the impact of an old car go before coming close to matching the impact of the manufacture of a new car?


Don't forget that the market for used cars was squeezed by restricted supply, driving many buyers toward new vehicles (and American-made ones at that) in an economic climate that would normally shift things the other way.


I get where you are going with this, but part of the benefit in my opinion was indeed to shift the timing of the purchases. I don't know the full analysis behind Edmunds' report, but durable goods purchses can be deferred a long time. In the case of cars, you can theoretially defer the purchase as long as you want to maintain the car and it passes the local safety/emissions regulations.

I don't know how Edmunds could calculate the amount of cars people "would have bought anyway" due to the lack of an accurate historical comparison to the current economic and unemployment situation. There seems to be general agreement this is the worst economic situation since the 1930's - where automotive transport was not quite so popular.