Blog Post

Transparency Needed in Trump’s Plan to Partially Repeal PHMSA’s Tank Car Rule

Jul 20, 2018 | Alan J. Krupnick, Justine Huetteman, Arthur G. Fraas

In March 2017, President Trump issued an executive order to reduce the regulatory burden on the energy sector. In response to this new priority, researchers at RFF undertook an examination of the costs and benefits of repealing or modifying oil and gas regulations, including analyses of the Bureau of Land Management’s methane rule, the Environmental Protection Agency’s methane rule, and the Bureau of Safety and Environmental Enforcement’s well control rule. In a new report, we look at the Pipeline and Hazardous Materials Safety Administration’s (PHMSA’s) 2015 rule to enhance the safety of tank cars transporting crude oil and ethanol (hereafter referred to as the tank car rule), which has been targeted for partial repeal by the Trump administration. The rule was promulgated primarily due to the increased use of rail for transport of oil and ethanol. Its promulgation also follows a major derailment of tank cars carrying crude oil that occurred in 2013 at Lac-Mégantic, Quebec, which resulted in 47 deaths and the evacuation of more than 2,000 residents. The damages from this incident were estimated to be more than $1 billion.

The regulatory impact analysis (RIA) accompanying PHMSA’s final tank car rule estimated a present value of net benefits of between -$1.7 billion and $1.3 billion over a 20-year period. The estimate of net costs (or negative net benefits) accounts for benefits only from avoided minor derailments (the RIA refers to these as low-consequence events), whereas the estimate of positive net benefits accounts for benefits from avoided minor and major derailments. Benefits from minor derailments include avoided environmental damage, clean-up costs, and a small number of injuries and/or fatalities; benefits from major derailments (the RIA refers to these as high-consequence events) are similar to minor derailments, but on a much larger scale, as in the Lac-Mégantic derailment.

To do a cost-benefit analysis of repeal, a number of steps must be taken to adjust the original analysis, such as accounting for sunk costs and benefits of the rule since its implementation began. We find that repealing this rule would yield net benefits to society of $1.5 billion considering only minor derailments but would yield net costs of $859 million considering both minor and major derailments. We maintain the net benefits range PHMSA used in its original RIA (i.e., we include two estimates, one for just minor derailments and one for both major and minor derailments). However, we believe it is prudent to factor in the potential avoided damages from major derailments due to uncertainty about the severity of derailments depending on location and scale, the increased risk of a derailment due to the increase in the use of rail for oil and ethanol transport, and the promulgation of this rule following a costly and deadly major derailment. 

But reliably calculating the benefits from avoiding major derailments is not an easy task. Because only one incident of that scale has occurred over the last two decades, PHMSA used a Monte Carlo simulation (which models the probability of different outcomes of an event when uncertainty is high and many variables contribute to an outcome) to estimate damages from these types of derailments.  The agency varied inputs like the number of derailments, the number of fatalities, the population density of the surrounding area, and wetlands damages. This type of analysis yields a probability distribution of damages, with the agency choosing the 95th percentile estimate to value these benefits. This means that there is only a 5 percent chance that the cost of damages from a major derailment will be higher than the value of the benefits. We found that the point at which the cost savings equal the forgone benefits of repeal of this rule is the 65th percentile damages. Above this percentile, the cost of damages rise steeply. The selection of damages along the probability distribution depends on the degree of risk aversion on the part of PHMSA.

In December 2017, the Trump administration announced its intention to repeal only the tank car rule’s braking provision, which requires installation of electronically-controlled pneumatic braking systems on certain types of trains (those of a certain length and carrying a certain class of flammable material). Mandated by the 2015 Fixing America’s Surface Transportation Act, the review of this requirement incorporated analysis by the US Government Accountability Office (GAO) conducted in 2016 and 2017. These GAO analyses recommended updating data inputs using more recent data, as certain inputs may have been overestimated in the original RIA, and including ranges of estimates due to uncertainty. The Trump administration released an updated RIA of just the braking provision last fall, showing that the costs of the provision were far greater than its benefits. However, this finding was reached by defining the damage from a major derailment as the mean estimate of damages from the original RIA’s Monte Carlo simulation instead of the 95th percentile estimate (as was used under the Obama administration). The mean estimate of damages from the Monte Carlo simulation is about one-fifth of the 95th percentile estimate.

In our report on the tank car rule, we found that the provision has net benefits of over $500 million when using the 95th percentile estimate of major derailment damages. In addition, in a separate scenario we account for sunk costs and benefits of the provision—which we believe to be best practice and which the Trump administration did not do in its updated RIA—and found that society would benefit on net from retaining the braking provision, even when we use the mean estimate of damages as the Trump administration did. PHMSA should revise its benefit and cost estimates in the updated RIA to reflect sunk costs and the associated benefits of actions already taken under the braking provision by the railroad industry. In addition, PHMSA should provide a range of damage estimates from the mean estimate to the 95th percentile. By adopting this approach, PHMSA could then fully and transparently explain its proposed choices.

The views expressed in RFF blog posts are those of the authors and should not be attributed to Resources for the Future.