Since their establishment in 1975, Corporate Average Fuel Economy (CAFE) standards have incited concerns about their effect on new vehicle sales. The standards, which require vehicle manufacturers to sell vehicles that achieve an average minimum level of fuel economy, are typically expected to reduce new vehicle sales. This is a concern, because the United States automobile industry currently employs about 4.3 million people (2.5 percent of the labor force), and any sales reduction likely leads to fewer jobs in the auto sector. Moreover, the CAFE standard’s effect on sales has unintended implications for the fuel savings and reductions in greenhouse gas emissions attributable to the standards. As the standards make new vehicles more expensive, some households opt to hold on to their less fuel efficient used vehicle longer, what is known in the literature as the “Gruenspecht effect.” This effect has been shown to be empirically significant, motivating its inclusion in analyses of CAFE standards.
The Gruenspecht effect was at the center of the most recent analysis of the Trump administration’s proposal—the Safe Affordable Fuel Efficient Vehicle Rule—to roll back the 2021–2025 standards to 2020 levels. The Environmental Protection Agency (EPA) and the National Highway Traffic and Safety Administration (NHTSA) recently published a preliminary analysis of the proposal and found that the rollback will reduce traffic fatalities. This effect stems from the impact the agencies claim the rollback will have on new vehicle sales: the rollback is anticipated to increase new vehicle sales (i.e., safer vehicles) and reduce the size of the entire (new and used) vehicle fleet, decreasing the number of accidents. However, this conclusion may be inaccurate due to flawed methodology employed in the agencies’ analysis.
In a recently published paper, my coauthors and I provide an ideal modeling framework that is underpinned by a basic economic model of consumer choice. Discrete choice models are the standard methodology for characterizing vehicle purchase behavior, and they can predict consumer choices under different policies. Importantly, they are used to predict the decision to purchase one product over another, such as the decision to own a used vehicle instead of a new vehicle. CAFE standards only regulate new cars and trucks. So, how do the standards affect the decision to buy a new vehicle? By increasing new vehicle prices, the standards are expected to decrease sales of new vehicles. But the magnitude of this effect depends on a buyer’s willingness to forego the purchase of a new vehicle. How much do new vehicle buyers value owning a new vehicle as opposed to a used vehicle?
In a working paper published today, I develop a new method for modeling this preference using something called “second choice data.” These data are based on survey responses to the following question posed to new vehicle buyers: “If the vehicle you bought did not exist, what vehicle would you have bought instead?” In my paper, I use responses from about 140,000 new car buyers in 2015 from the New Vehicle Customer Survey (NVCS), administered by MaritzCX. The survey collects a series of responses about the second-choice vehicle, including the make, model, trim, engine size, body style, fuel type, model year, and—importantly for the current context—whether the vehicle would be bought new or used. These data provide a sense of how willing new vehicle buyers are to substitute to a used vehicle. For example, if most respondents state that they would have bought a used vehicle instead of their new vehicle, we would expect that moderate, undesirable changes to the characteristics of new vehicles (such as an increase in price) would cause many consumers to buy a used vehicle instead of a new vehicle.
It turns out that a vast majority (92 percent) of new vehicle buyers surveyed stated that they would have bought a different new vehicle had their first choice vehicle not been available. Considering that only about one-third of all vehicle purchases in 2015 were new vehicles (with the remaining two-thirds being used), this suggests that new-vehicle buyers share a strong value for owning a new vehicle relative to the average vehicle buyer. Therefore, we can expect a policy like CAFE standards to have a modest effect on new vehicle sales.
The results of my simulation suggest that a one-percent tightening of CAFE standards would reduce new vehicle sales by 0.02 to 0.08 percent, depending on the assumptions made regarding technology costs and consumer valuation of fuel cost savings. This range represents a modest effect of CAFE on new vehicle sales, and is similar in magnitude to recent work by RFF Senior Fellow Joshua Linn (which is based on a different methodology).
A key feature of the method I develop is that it is relatively straightforward to adopt and use for analyzing policies like CAFE. The EPA, NHTSA, and other federal and state agencies typically have tight deadlines for analyzing changes to energy and environmental policies like CAFE, so this feature may prove valuable to them if they decide to incorporate a consumer choice model in their future analyses.
