To justify rolling back the U.S. LDV efficiency standards, NHTSA and EPA put a thumb on the cost-benefit scale

The Trump administration’s bid to roll back the U.S. light-duty vehicle efficiency standards depends, essentially, on reversing the conclusions of the benefit-cost analysis that the Environmental Protection Agency and the National Highway Traffic Safety Administration did just two years earlier. And the administration has to accomplish that using the same agencies, the same data, the same research literature. Not an easy lift.

How do they do it? As we explain in a just-published briefing, by changing underlying assumptions, data inputs, and models used to make projections. In plainer English: NHTSA and EPA put a thumb on the scales used to weigh benefits against costs.

You should read the briefing, of course. But here’s the gist.

The United States has two separate but harmonized light-duty vehicle standards that regulate fuel efficiency (NHTSA’s purview) and greenhouse gas emissions (EPA’s). Each agency performs its own benefit-cost analysis, which involves projecting technology availability, costs, consumer fuel savings, and other factors influencing the final effect of the regulation. Each uses its own computer model. While the models don’t produce identical projections, from 2009 until 2018 they were closely aligned. But the analysis for the Trump Administration’s rollback proposal, mostly done by NHTSA using its Volpe Model, diverges sharply. And curious as it is that the 2018 rollback contradicts prior years’ results, that the model EPA has spent years developing and continues updating was not even used for the rollback proposal is even curiouser.

The table below summarizes three different benefit-cost estimates done for the U.S. LDV efficiency regulations. EPA’s January 2017 Final Determination is the one that the Trump administration’s proposal would reverse. NHTSA’s estimate for the 2016 Technical Assessment Report, the last NHTSA analysis of the Obama Administration, is the estimate most directly comparable to the benefit-cost analysis for the 2018 proposal. 

U.S. fuel economy and GHG regulation impact estimates

  Factors EPA Final Determination (January 2017) NHTSA Technical Assessment Report (July 2016) NHTSA Proposed Regulation (August 2018)
    Cost Benefit Cost Benefit Cost Benefit
Societal Impact ($ billions) Technology Cost $33   $87   $253  
Crashes   $1.8 $1.2   $197  
Congestion $6.2   $5.0   $52  
Fuel Savings   $92   $120   $133
Pollution   $29   $37   $6
Mobility   $10   $9   $61
Other Impacts   $12   $16   $126
Overall effect Model years affected 2022-2025 2022-2028 2021-2029
Total number of vehicles 65 million 11.5 million 160 million
Benefit-cost ratio 3.7-to-1 2.0-to-1 0.6-to-1

The chart below shows the impact estimates of the table above on a per-vehicle basis. So, for example, the EPA in 2017 estimated $600 in total average per-vehicle cost and $2,200 in benefits; the three largest factors in EPA’s analysis are the technology cost ($500 per vehicle), fuel savings ($1,400), and pollution benefit ($440); and the overall benefit-cost ratio is 3.7 to 1.

U.S. fuel economy and GHG regulation impact estimates, per vehicle
U.S. fuel economy and GHG regulation impact estimates, per vehicle

This table summarizes the changes in estimated impacts from the earlier analyses to the Trump administration proposal: higher technology costs (2x–3x the earlier analyses), fatalities and crash costs (as much as 120x increase), and congestion costs (3x–7x); lower fuel-saving benefits (by 21%–41%) and pollution benefits (by 89%–92%).

Per-vehicle efficiency and GHG regulation impacts from three analyses

Regulation impact Impact per regulated vehicle ($/vehicle Change, 2016/2017 to 2018 analysis Factors in Change
  EPA 2017 NHTSA 2016 NHTSA 2018    
Technology Cost -$502 -$758 -$1,581 109% to 215% Less technology available
Technologies offer less benefits
Technology costs greater
Crash $28 -$10 -$1,236 -4553% to -11937% Sales response: Lower vehicle sales (by about 1%) and more use of older vehicles
Rebound effect: Drivers capitalize on fuel savings by driving more
Congestion -$96 -$44 -$324 238% to 644% Rebound effect
Fuel savings $1,409 $1,048 $832 -21% to -41% In absence of new 2020+ standards, efficiency assumed to increase from 36 in 2020 to 38.4 mpg in 2026
Pollution $441 $323 $35 -89% to -92% CO2 damages reduced by 85%
Rebound effect
Mobility $155 $79 $382 146% to 383% Rebound effect
Other impacts $187 $139 $790 322% to 469% Largely to offset crash impact, as drivers freely choose to drive more

In our briefing, we peel back the onion of this re-analysis in detail, and show exactly how ploys like restricting technology availability, increasing the cost of existing technologies, assuming vehicle sales decline due to efficiency improvements, increasing fatalities from greater use of older vehicles, and removing energy and emission-reduction benefits combine to inflate costs by $300 billion and devalue benefits by $100 billion. (For even more detail, see our public comments here.) For illustrative purposes, it’s useful to just look at how the benefit-cost ratio changes when the assumptions are changed to more reasonable values.

The chart below estimates the impact of the technical assumptions within the agencies’ 2018 benefit-cost analysis. Starting from the left, the Trump administration analysis resulted in an estimate that showed the augural 2025 standards providing a net disbenefit of -$176 billion (i.e., $502 billion cost against $326 billion in benefits). Moving to the right, each step with a light green box changes one major assumption at a time. These steps show our best research-based estimates on how the assumption changes individually impact the regulation’s benefit-cost analysis. The steps are cumulative. The final column to the right includes all changes.

Net societal impact after revising 2018 agency modeling assumptions
Net societal impact after revising 2018 agency modeling assumptions

The first step changes technology cost to reflect the most recent data on vehicle efficiency technology cost and availability (in year 2025, this drops the incremental cost by two-thirds). That reduces costs in the analysis by $170 billion. The next most consequential 2018 modification is the agencies’ decision to make vehicle sales decline in consequence of post-2020 efficiency improvements. This novel sales-response assumption adds $110 billion in economic costs due to increased crash-related costs. Following these, in order of their impact on the final outcome, are pollution benefits ($46 billion), fuel-saving benefit ($42 billion), and the changes to the rebound effect ($29 billion).

The cumulative effect of the changes in the agencies’ 2018 analysis depicted in chart is the difference between a net cost of $176 billion and a net benefit of $221 billion—a positive swing of $397 billion. The 2018 analysis adds about $300 billion in technology, crash-related, and congestion costs, while devaluing about $100 billion in fuel savings and pollution benefits.

The agencies justify fuel economy and greenhouse gas regulations when the benefits to society outweigh the costs of regulation. By the use of numerous modeling tricks and faulty assumptions, the 2018 proposal claims fuel economy standards have higher costs than benefits—the benefit-to-cost ratio is 0.6 to 1—thereby justifying a rollback. We estimate that fixing these issues would flip the outcome: benefits would outweigh costs by about 3.2 to 1. Unsurprisingly, that net-positive result falls neatly in line with all cost-benefit analyses conducted by the agencies from 2009 to 2017, supported using two different modeling approaches. Could it really be that in just one year, innovation has stagnated to such an extent that fuel-efficiency regulations are no longer a net benefit to consumers and society? Not likely. The evidence suggests instead that the agencies deliberately tipped the scales with NHTSA’s Volpe model to satisfy the Trump administration’s goals of deregulating industry and crippling efforts to address the climate crisis through U.S. public policy.