As the US Environmental Protection Agency begins enforcing methane fees on the oil and gas industry, an important question begs answering: How can regulators best calculate the true extent of methane leaks?
Economists love the idea of carbon pricing, so it’s a big deal that the Inflation Reduction Act (IRA) introduces the first national price on greenhouse gas emissions: a charge that phases in to $1,500 per ton of methane emitted by the US oil and gas industry, the implementation of which EPA recently pushed forward. That dollar value corresponds to the 2021 interim federal estimate of the social cost of methane (and EPA’s recent update suggests a similar estimate of $1,600). This sounds like a victory for carbon pricing, but the devil is in the implementation details, as the scientific advisory board to the US Environmental Protection Agency (EPA) pointed out in a letter to the agency in mid-January. As the saying goes, what gets measured gets managed—and that’s particularly true here.
In standard economic theory, a carbon price creates a private incentive to reduce pollution: if I cut my methane emissions by one ton, I save $1,500, creating a profit motive to undertake any efforts to reduce emissions that cost less than $1,500 per ton. But what if the measurement is off when my taxes are calculated? The IRA specifies that the methane fee should be calculated using the emissions tallies from the EPA Greenhouse Gas Reporting Program. But it is well established that the reporting program drastically undercounts actual methane emissions from the oil and gas industry—perhaps by a factor of four. Poor accounting thus can blunt the incentive that the methane fee intends to create.
This potential for inaccurate calculations is not news to the drafters of the IRA, of course, which is why the section of that law containing the methane fee also directs EPA to update how the agency calculates methane emissions in the oil and gas industry, to make sure that the numbers “are based on empirical data … [and] accurately reflect the total methane emissions” from the covered facilities. Measuring methane emissions is particularly challenging, because a disproportionate amount of methane emissions from oil and gas infrastructure comes from intermittent “super-emitters.” These super-emitter events are large, typically unintended, episodic leaks such as those due to malfunctioning equipment. Since these unintended leaks are difficult to measure and often go unnoticed, super-emitter events are not tracked routinely nor included in the Greenhouse Gas Reporting Program.
Enter EPA’s recent proposal for how to quantify emissions from these super-emitters in the reporting program. The idea is as follows: suppose a major methane leak is identified at the site of an oil well, either during an inspection or through remote sensing like an aerial flyover. Further suppose that the observed leak rate is 200 kilograms (kg) of methane per hour, but we don’t know how long the oil well has been leaking. EPA proposes that we should assume that methane has been leaking at that rate since the last time the well was observed to be emitting less than EPA’s super-emitter threshold (set at 100 kg per hour). In many cases, this date would be the date of the last inspection at the well that confirmed emissions under the threshold or, if no related record exists, EPA suggests using a default of 182 days (six months). That duration is the typical time between inspections, assuming the operator is properly conducting the required twice-annual inspections, as will soon be required under EPA’s separate methane regulations.
This all sounds reasonable enough, but the EPA scientific advisory board’s recent critique points out a flaw with this approach: these calculations could simultaneously underestimate emissions on average while vastly overestimating emissions from the individual leaks that are caught.
Why? Well, first, the scientific advisory board points out that most major leak events have durations of less than a day, citing recent research on the issue, whereas the duration assumed in the EPA proposal generally would be much longer than that. If a major leak is detected at a random time during the 182-day window between those twice-annual inspections, then on average, the time elapsed since the last inspection would be 91 days (half of 182). Thus, when an individual leak is caught, EPA’s proposal would imply using a much longer duration (91 days on average) than is realistic. But at the same time, the intermittent nature of these leaks means that it is possible that most leaks would go completely uncounted; say, because no one happened to be checking when the leak was happening.
Credit: noomcpk / Shutterstock
A simple stylized example illustrates the point. Imagine an oil well that’s not leaking any methane 99% of the time, but a leak occurs 1% of the time at a rate of 500 kg per hour. Actual emissions over a 182-day window would be ~22,000 kg ((99% × 0 + 1% × 500 kg per hour) × 24 hours × 182 days). If we could enforce the methane fee based on that actual value, the appropriate charge would be $33,000 (22 metric tons of methane × $1,500 per ton). (A caveat: for simplicity, the fee calculations shown here do not account for the thresholds under which methane fees are waived.)
If we randomly inspect this oil well once during this 182-day window, we have a 1% chance of observing it during the super-emitting period and would estimate emissions at ~1,100,000 kg on average (500 kg per hour × 24 hours × 91 days, where 91 days is the average duration since the last inspection noted above). The methane fee as calculated would be ~$1.6 million (1,100 metric tons × $1,500). This calculation represents a 50-fold overstatement of both actual emissions and the appropriate fee. On the other hand, we have a 99% chance of missing the leak completely, treating it at 0 kg, and charging no methane fee whatsoever, which represents a 100% understatement. In expectation, our calculation comes out to ~11,000 kg (1% × 1,100,000 kg + 99% × 0 kg), which is 50% lower than the actual emissions from this well.
This result also has real consequences for operators’ incentives to abate methane leaks. In our stylized example, the operator pays 50 times too much when the leak is caught, but has only a 1% chance of actually facing that outcome. The net result cuts the incentive to abate by half the $1,500 face value of the methane fee, substantially reducing the effective incentive of the fee for mitigation.
These large default durations also create other diverging incentives for operators. On the one hand, the possibility of an excessive charge could encourage better and more frequent monitoring, which would reduce the assumed leak duration when the fees are calculated. On the other hand, high default durations could discourage the detection and reporting of leaks at all, since doing so could trigger very large fees. This potential for evasion emphasizes the importance of EPA’s new super-emitter program, established as part of the agency’s recently finalized methane regulations, which allows certified third parties to identify and report leaks—not just the operators themselves.
The above stylized example is for a single oil well, but the analogy can work across wells and operators, too. The implication is that, although using this calculation method to assess methane fees for super-emitting events may yield results that are too low on average across operators, operators that are caught for super-emitting events may end up being charged for more—and potentially a lot more—than their actual emissions. This example highlights the tricky issues around designing proper pricing incentives for externalities that are difficult to measure and attribute to individual operators. Future improvements in remote sensing, such as MethaneSAT, offer potential for improving greenhouse gas accounting and facilitating effective carbon pricing. But until then, the wonky and technical methods for emissions accounting may start to have major consequences for the effects, and effectiveness, of the first nationwide carbon price in the United States.
