The federal government employs people of many different professional backgrounds, but among them you will not find fortunetellers. Yet scientists, analysts, and lawyers could sometimes use an assist from seers in completing the work they are asked to do.  Conducting cost-benefit analyses of agencies’ proposed regulations necessitates predicting their impact on the economy, an unfathomably complex institution composed of choices made by millions of actors around the globe. Accurately projecting how it might respond to proposed regulatory changes requires assumptions about how people and companies will respond. This complexity and uncertainty are among the reasons why agencies should proceed with caution when establishing new rules.

Through its responsibility to implement the Clean Air Act, the EPA regularly issues economically significant rules.  While these rules restricting emissions will certainly generate substantial costs, reducing pollution is a goal sufficiently worthwhile to justify them, within reason.  However, if companies and individuals are forced to bear these costs, it is critical that the promised reductions in emissions occur.

A working paper produced by National Bureau of Economic Research economists Mark R. Jacobsen and Arthur A. van Benthem suggests these reductions may not fully occur in the case of the CAFE regulation issued last October. The authors call the effectiveness of CAFE into question based upon a concept known as the Gruenspecht effect. When fuel efficiency standards are raised, the prices of new vehicles will inevitably increase (by around $1,800 for this rule, according to the EPA). Since used vehicles are a substitute for new vehicles, as the price of new vehicles rises, some consumers may opt to buy used vehicles, increasing used vehicle demand, raising their value.  and therefore reducing the likelihood of vehicles being scrapped.  This effect, coupled with the possibility that people may continue to drive their current vehicle rather than pay more for an update, leads to longer lives for the vehicles currently on the road, delaying the adoption of new vehicles subject to the higher efficiency standards. Therefore, by changing the composition of the vehicle fleet, fuel economy regulation may not produce reductions as large as they would be if the number of new vehicles purchased and used vehicles scrapped was unaffected. The study estimates the magnitude of this effect at 13-23 percent, depending upon the sensitivity of the used vehicle scrap rate to changes in price.

It is difficult to translate this finding into an exact estimate of the error in the EPA cost-benefit analysis because the two analyses look at different data sets – the EPA considers lifetime savings from new vehicle sales, while the NBER estimates the fuel savings within the entire U.S. fleet. Furthermore, the Gruenspecht effect is based upon changes in relative price between new and used vehicles due to changing regulations.  However, the EPA estimates that these increased fuel economy standards will save drivers $3,400 to $5,000 in fuel costs over the lifetime of the vehicle, despite the higher sticker price faced at the time of purchase. Therefore, if consumers are aware of the future savings, they may be willing to bear the cost of new cars, and used vehicles may not become dramatically more attractive.

Yet these individual fuel savings raise other questions about agency rulemaking. By relying on these private savings to justify the costs of the rule, rather than doing so solely through the public benefit of less pollution, the EPA assumes the position of imposing certain preferences on consumers  (in this case, that Americans opt for more fuel-efficient cars, even though some may not make this selection freely despite the potential savings). Yes, the private savings from reduced fuel consumption are connected to the environmental benefit of limited emissions, but if regulators are justifying the costs of the rule with private savings, rather than considering them a side benefit, it implies that those savings are part of the end goal of the rule.

The challenges naturally surrounding predicting and explaining consumer behavior demand caution and thorough analysis from regulators. Perhaps it is better for individuals to choose more fuel-efficient cars to save themselves money over the long haul. But then the question becomes why government must make this decision for them. If it was preferable for all consumers, self-interest should naturally guide them to it.  If there is an information problem, government ought to alleviate it to allow better choices. But if not, could there be something else about car decisions besides fuel economy that regulators are overlooking? The NBER’s finding that people will choose to keep their used cars suggests that not everyone agrees with the “items on the menu” in the EPA’s “CAFE.”