In fragile economic times, the U.S. government occasionally enacts stimulus programs to provide a jolt to consumer spending. This was certainly the case during the Great Recession, when the government utilized both explicit stimulus programs, such as the American Recovery and Reinvestment Act of 2009, and an expansion of federal credit programs to generate a similar effect. When initiatives such as those induce the intended stimulus, they are generally (though not universally) heralded as successes. But for future policymaking, a question more important than if a program worked is why it worked.
A new study by GCFP co-director Jonathan Parker explores a smaller but impactful stimulus program from the Great Recession and the factors contributing to its unexpectedly high participation rate. Colloquially known as “Cash for Clunkers,” the Car Allowance Rebate System (CARS) program provided instant rebates from the federal government of $3,500 or $4,500 to consumers who traded in their old, gas-guzzling vehicles for newer, more fuel-efficient ones in July and August of 2009.
The program’s take-up was tremendous, with nearly seven times more transactions submitted than anticipated. In their paper, “Accelerator or Brake? Cash for Clunkers, Household Liquidity and Aggregate Demand,” Parker and coauthors Daniel Green (MIT), Brian Melzer (Northwestern) and Arcenis Rojas (Bureau of Labor Statistics) argue that the remarkable take-up of the program was due in large part to the liquidity provided by the program in addition to the price subsidy offered. Unlike a concurrent stimulus program that subsidized first-time home purchases through year-end tax credits, CARS gave consumers immediate cash with which they could finance the purchase of a new vehicle.
Because the CARS program had strict eligibility criteria for “clunkers,” such as age and fuel efficiency, the authors are able to identify the effect of the program on new car purchases by comparing owners of “clunkers” with owners of “close-to-clunkers” during the program period. Overall, they find that CARS caused roughly 506,000 new vehicle purchases, worth $11 billion, in the third quarter of 2009 at a fiscal cost of $3 billion. They also measure how the response differed with the economic subsidy provided by the program. The economic subsidy is not the value of the rebate credit, but instead is the rebate credit (either $3,500 or $4,500, depending on how much more fuel efficient one’s new vehicle is than her old one) minus the market trade-in value of the clunker. Parker and his co-authors show that CARS raised the probability of purchasing or leasing a new vehicle by half a percentage point per $1,000 of economic subsidy.
Having established that CARS had an effect on new vehicle purchases during the program period, Parker et al. shift their focus to the cause of the program’s high participation rate. The authors show that the high take-up of the CARS program was driven by the provision of liquidity along with the economic subsidy. In particular, participation in the program was less likely when, despite the economic subsidy, the program provided little liquidity. How do the authors show this? The maximum rebate amount of $4,500 (claimed by 71% of CARS program participants) actually exceeded the total down payment made on nearly 70% of new vehicle purchases during the program period. Thus most households with clunkers could participate in the CARS program using only the CARS rebate and so not dipping into their own savings. However, some households could not use the CARS rebate this way, specifically households that had outstanding loans on their clunkers. Loans on the clunker would have to be paid off immediately upon trade-in, so that these households would have to cover any difference between the CARS subsidy and their down payment from their own pockets.
The authors show that the participation rate was nearly zero for households owning clunkers that were securing vehicle loans. The large response to the program came almost entirely from households that could use the CARS rebate as a down payment, and not from households that had to first pay off an existing loan. Importantly, there is no lower response for households with debts that are not secured by their clunker. Households with unsecured debts participated in CARS at similar levels to households without. Nor is the lower response of households with clunkers that secure loans a spurious relationship due to differences in household income, liquid assets, or the size of the subsidy between households with and without vehicle loans.
In sum, liquidity fueled the massive response to CARS. Had the CARS rebate been something that one could claim on a tax return rather than a payment at the car dealer, this price subsidy alone, without the benefit of liquidity, would have generated a much smaller response to the program.
The paper offers a number of insights for future stimulus programs. The impact of stimulus programs like this can be limited by household liquidity constraints, particularly during recessions when financial constraints are most binding. Programs that combine liquidity and subsidies, such as by issuing rebates immediately at the time of purchase rather than later, not only will maximize participation but will also be more equitable across strata of household liquid savings. When rebates provide sufficient liquidity, even low-income households seem to be able to obtain financing from private markets, at least for goods that can be used as collateral. Finally, in crafting future stimulus programs, policymakers would be wise to strongly consider the element of liquidity in order to achieve their program’s objectives.