It has been a year since the government's auto stimulus plan commonly known as Cash for Clunkers. Coming upon this anniversary of sorts, we have started getting calls from media wanting to know if, now with the benefit of hindsight, our analysis has changed.
Edmunds.com watchers will remember that our assessment last year stirred up quite a ruckus. My Op Ed in the WSJ was debated on the Senate floor and triggered an uncharacteristicly apoplectic post on the White House blog (where we were accused of counting car sales on Mars.) A major car dealer accused our analysts of smoking medical marijuana.
All this because we put numbers to what economists already knew: That influencing the timing of consumer durable purchases is easy. Creating new purchases is not. This means that incentive programs always suffer from a high cost of incrementality (sales that wouldn't have occurred soon anyway.) Supporting this, one noted economist weighed in here.
Last year, sales were delayed in April, May, June and early July by consumers waiting for the program to launch. (This is why car companies never pre-announce an incentive program.) Sales then soared during the first and second waves of funding, only to collapse as the program ended. It took months for car sales to return to their baseline rate.
Proponents love to point to the speed at which the funding ran out, the fact that 2009 model year inventory was cleared off dealer lots and the that consumers trading in "clunkers" loved it. Indeed Secretary of Transportation Ray LaHood proclaimed C4C "the stimulus program that really worked."
Skeptics fret about the duration (even confirmed Keynesians would point out the program should have lasted much longer), the timing (why start when sales are due to spike anyway during the summer selling season), or even pesky issues such as the fact that inventory shortages soon developed, pushing up prices up for all car buyers.
All we offered up was some math, tracking sales prior and during the program and then forecasting sales after. This pointed to a net increase is sales of around 125,000. The program ended costing a shade over $2.8B, putting the incremental price tag at north of $24,000 per extra sales. (Uncomfortably, this was higher than the price of the typical vehicle sold.)
Like I said, incrementality is expensive.
With all the inquiries we have been receiving, I asked our analysts to take another look -- this time at actual sales, not a forecast. In fact, to play it safe, they looked at the numbers several ways. The results vary, but not by much. One put the incremental sales at 135,000, the other at 101,000. Even in the bast case estimate, the incremental costs was still above $20,000.