Jeremy West

Implemented in the midst of the 2009 recession, the U.S. Cash for Clunkers program aimed to boost sales in the struggling automobile industry. Eligible households were provided with subsidies when they scrapped their old “clunkers” and purchased a new vehicle. The argument was that this would shift expenditures “…from future periods when the economy is likely to be stronger, to the present…” (Romer and Carroll, 2010). However, to serve national energy and environmental goals the policy layered on a second requirement, that the new vehicles be of sufficiently high fuel economy. A recent CEEPR working paper by Hoekstra, Puller, and West (HPW, 2015) finds that this multifaceted program design actually caused Cash for Clunkers to reduce overall revenues to the industry the policy was designed to help.


Cash for Clunkers as a Stimulus Policy

The academic and policy spheres have seen significant debate regarding the merits of various federal policies aimed at stimulating the economy during the last few recessions. In 2009, with the international automobile industry floundering, policies to stimulate new vehicle sales seemed particularly promising. Indeed, more than 15 countries implemented programs similar to Cash for Clunkers to target new vehicle sales. 

In the United States, federal policymakers constructed the Car Allowance Rebate System. For nearly two months during the summer of 2009, households who scrapped an eligible vehicle were subsidized up to $4500 towards the purchase of a new car or truck, provided the purchased automobile met certain fuel economy conditions. By designing the policy in this way, policymakers hoped to meet two objectives: the program would provide immediate stimulus to the struggling automobile industry, and it would help reduce American use of gas-guzzling vehicles that contribute to climate change and local air pollution.


Evaluating the Effects of Cash for Clunkers on Automobile Sales

Cash for Clunkers was designed to affect not only the timing of households’ new automobile purchases but also the composition of new vehicles purchased. HPW provide causal evidence on how the program affected both dimensions. They exploit the program’s discrete eligibility cutoff to obtain a compelling counterfactual for the timing and type of new vehicle purchases made by subsidized households. Specifically, the program’s eligibility cutoff of 18 miles per gallon serves as a natural experiment in program participation: households who owned a clunker rated at 18 MPG or lower could receive the subsidy, whereas those with a clunker at 19 MPG or higher were ineligible for subsidies. 

The authors use this approach with administrative data on household vehicle ownership and purchases from the Texas Department of Motor Vehicles. First, they show that households who were “barely eligible” appear very similar to “barely ineligible” households in terms of their pre-program vehicle fleets and other household characteristics. Then, they estimate the counterfactual purchase timing for subsidized purchases; the results indicate that about 60 percent of subsidies went to households who would have bought a new vehicle during the two months of the program anyway, with the remaining subsidies pulling purchases forward from across the following eight months or so. By ten months from the start of Cash for Clunkers, there is no difference in purchase probabilities for barely eligible relative to barely ineligible households.

Next, HPW use this same eligibility cutoff-based strategy to evaluate how Cash for Clunkers affected new vehicle spending, rather than just sales counts. They use the ten month time period starting with the program, which holds constant the probability of a household purchasing any new car or truck and allows them to identify just the differences in automobile characteristics. The estimates show that barely eligible households who purchased under the program spent an average of five thousand dollars less (transaction price) on new vehicles than did barely ineligible households who purchased a new vehicle during the same ten month period of time – that is, barely eligible households were incentivized to purchase slightly earlier and spend significantly less. Assuming a similar effect size outside of Texas, these results suggest that Cash for Clunkers actually reduced aggregate new vehicle spending by around three billion dollars nationwide over a period of less than a year.

HPW show that these differences in vehicle expenditures are explained by the generally negative relationship between automobiles’ fuel economy and price, as shown in Figure 1. To meet the fuel economy restrictions of Cash for Clunkers, subsidized households purchased vehicles that were higher fuel economy, but also lower performance, smaller size, and lower book value. 

This Cash for Clunkers experience has implications for future policymaking. While hindsight is always 20/20, it certainly appears that the primary policy goal – stimulating revenues to the auto industry – was undermined by the “add-on” energy and environmental objective of the policy. Dual policy goals, even those that are individually worthy, can sharply undermine each other when implemented as a single policy.

Christina Romer and Christopher Carroll. Did Cash-for-Clunkers work as intended? White House commentary (2010). Council of Economic Advisers.
Mark Hoekstra, Steven L. Puller, and Jeremy West (2015), “Cash for Corollas: When Stimulus Reduces Spending.” CEEPR WP-2015-005, MIT, April 2015.