Air Pollution Co-Benefits Matter
Ignoring health benefits of air quality improvements costs everyone
We’re all suffering from post-election exhaustion. As we slip back into business-as-usual, it’s tempting to pay less attention to the politics of everything. But there’s some high-stakes maneuvering that we should be following. This includes efforts to knock some teeth out of federal air pollution regulations.
This campaign is advancing on multiple fronts. I want to focus on two in particular. First, a proposed reform of cost-benefit analysis protocols which aims to discount “co-benefits” of air pollution regulation. Second, a move to change how health-based air quality standards get set. Taken together, the sum of these developments is even more concerning than the individual parts.
Discounting “Co-benefits” in cost-benefit analysis
For decades, Republican and Democratic administrations have agreed that costs and benefits should be considered in the rulemaking process. The EPA has a statutory obligation to estimate the costs and benefits of major regulations. Cost-benefit analysis cannot be used to directly determine health-based standards for criteria pollutants, but it can dictate the choice between regulatory alternatives (see here for details).
If you want to dismantle a cost-effective environmental regulation, you need to find a way around these obligations. One strategy involves meddling with how benefits are defined and quantified. And that’s precisely where the EPA is heading.
We’ve blogged about some of these tactics already. But the accounting change I want to focus on involves the counting of “co-benefits”. These are benefits from pollution reductions (such as avoided premature deaths) that are caused by a regulation but associated with pollutants that are not directly targeted by the regulation.
Air pollution problems are interconnected (source)
Let’s think about this co-benefit idea for a minute. We’ve all made big decisions that impact our lives in lots of ways. Focusing narrowly on one objective when there are many important factors to consider can lead to some bad choices. Extend this rationale to environmental regulations, and it seems quite sensible to count all significant impacts of a regulation (good or bad) in cost-benefit analysis. This approach is not only rational, but it’s also totally consistent with law and historical precedent.
Under the Trump Administration, however, the EPA has been pushing an alternative approach that considers only the benefits from reductions in the pollutant that is directly targeted by a rulemaking. This seemingly small tweak could have surprisingly large health consequences.
Consider, for example, the Clean Power Plan (CPP) which was designed to reduce greenhouse gas emissions (GHGs) from the electric power sector. The graph below summarizes recently re-estimated costs and benefits reported by the Trump administration in an assessment which aims to repeal this Obama-era rule:
The graph summarizes annual costs and benefits projected for 2030 (3% discount). Estimates are taken from this Regulatory Impact Assessment (Table ES-9)
Total benefit re-estimates are less than a third as large as those estimated by the Obama administration for a number of reasons. But these changes notwithstanding, the CPP rule still sails through the cost-benefit test … so long as “co-benefits” associated with air quality improvements are counted.
Another striking example can be found in the “mercury rule” which also regulates emissions from power plants. Under the Obama administration, annual costs were estimated at $9.6 billion. Benefits from decreasing mercury were estimated at only $4-$6 million. Annual co-benefits (mainly from reductions in particulate matter pollution) were estimated at $37-$90 billion. This heavy reliance on co-benefits to cost-justify has been controversial. The current EPA has reportedly submitted a draft revision that removes co-benefits from the analysis (turning this net-winner into a net-loser).
Why are health co-benefits so darn big?
The Wall Street Journal will tell you this is just the EPA “juking the numbers”. Other critics suggest that co-benefits accounting is double counting. But this is wrong. These co-benefits estimates include only the incremental benefits caused by the rule being evaluated, over and above the expected benefits generated by other rules on the books.
In fact, the outsized role of co-benefits can be traced back to how we regulate tiny and pernicious particulate matter pollution (known as PM2.5).
Image courtesy of the US EPA
In the above examples, a majority of the co-benefits are associated with reductions in PM2.5 exposure. Recall that these estimated health co-benefits (including estimates generated by the current administration) significantly exceed abatement costs. This suggests that regulations directly targeting PM2.5 are leaving lots of room for cost-effective improvements.
There are a number of reasons why current PM2.5 regulations might fall short.
- One has to do with the PM2.5 regulatory framework. For example, once national PM2.5 standards have been met, there’s little incentive to reduce exposure-levels further (even in places where the health benefits of exceeding the standard clearly exceed costs).
- Another has to do with regulatory lag. Today’s PM2.5 standards were based on the state of the science back in 2009 which had yet to clearly demonstrate the health benefits associated with lowering the standard below current levels.
We are now on the cusp of a review of federal PM2.5 standards. In preparation, the EPA has recently fulfilled its obligation to gather the most recent and relevant science on health impacts. The science has advanced considerably since 2009. Among the highlights: A growing body of evidence shows a relationship between PM 2.5 exposure and mortality at exposure levels well below (i.e. less than half) the current standard. And mounting evidence that health benefits from incremental reductions in PM exposure are increasing as air quality improves.
If you’re hoping that this expanded scientific understanding will lead to more stringent PM2.5 limits, don’t hold your breath. As part of a larger effort to “better separate scientific judgments from policy decisions”, the EPA recently abolished the science advisory panels that have informed standard setting since the 1970s. This leaves a seven-member committee of Trump administration appointees responsible for advising the EPA on what level of pollution exposure is safe enough.
A one-two punch to air quality improvements
If the revised PM2.5 standards fail to mobilize cost-effective opportunities to reduce harmful exposure, there will be health benefits ripe for the co-benefit picking. Achieving cost-effective reductions in particulate pollution via co-benefits may not be first-best. But given binding regulatory constraints, this can be a second-best approach worth taking. Proposed cost-benefit reform would fence off this approach, making it much harder to opportunistically pursue cost-effective health co-benefits.
In past years, federal environmental regulations have delivered significant and cost-effective improvements in our air quality. Rigorous consideration of the science, costs, and benefits have been guiding principles in this endeavor. A departure from these principles is not reform. It’s a move in the wrong direction.
Air quality has been improving.. let’s keep it that way!
Can I be critical of co-benefits, without implied endorsement of the rest of this EPA’s agenda?
1) The current PM2.5 regulations are ambient air quality standards, set nationally but overseen locally. So if the mercury rule or the Clean Power Plan reduces PM2.5 from power plants, in theory that frees up states to relax regulations on PM2.5 from other sources and still meet the ambient standard. That would yield zero co-benefits. The fact that “once national PM2.5 standards have been met, there’s little incentive to reduce exposure-levels further” is an argument against counting co-benefits, not in favor.
2) The mercury rule and the Clean Power Plan both apply the same measure of the benefit per ton of PM2.5 reduction. But if marginal abatement benefit curves are decreasing (like we draw them on blackboards in classrooms), then the more particulates are reduced, the smaller should be that benefit per ton of reduction. Each successive regulation that reduces co-benefits should be using a smaller benefit per ton, reducing the co-benefits.
What’s really going on here? Mercury impairs cognitive development in children around the world, a terrible problem that is difficult to monetize in a US-based cost-benefit analysis. The $6 million direct benefits of reducing mercury in the EPA’s analysis is their 2011 estimate of the damage from the lowered IQ of American children who eat the freshwater fish caught by parents who fish for fun. That’s comically, inappropriately small. Rather than relying on co-benefits to justify not poisoning the world’s children, we need to do a better job of assessing the real direct damage from global mercury than the EPA did in its 510 page analysis of the rule in 2011.
Great blog. I’m an avid reader. Keep up the good work.
Thanks for the terrific comments. You’ve touched on two great points that were not well-articulated in the blog.
Taking these points in reverse. On (2). There is increasing evidence that MB curves for PM2.5 are not your textbook Econ 101 curves. As air quality improves, we are seeing more studies of health impacts at levels below our current standards. Recent estimates present evidence that the concentration-response (wrt mortality) relationship might be strictly *concave*. Krewski et al 2009 is one important/early reference. But the recent ISA offers further evidence that the more particulates are reduced, the greater the marginal health benefits. This complicates the pictures we draw on the board… and our thinking about health-based thresholds!
On (1), great point. My arguments were too fast and loose. But to the extent that compliance w NAAQS is achieved via fixed and irreversible investments, one could plausibly argue additional improvements via co-benefits are unlikely to get undone in the short run (and this buys time for the regulatory lag to correct). Also, current standards don’t appear to bind in many counties, suggesting cost-effective co-benefits identified in these other rulemakings would be additional (versus offset). But you are exactly right that co-benefits are not an ideal way to compensate for a standard that is insufficiently stringent. And that it’s entirely possible that co-benefits will get offset so far as they relax a binding constraint.
I also agree that a compelling argument for the MATS rule is global. There’s a PNAS paper by Giang and Selin that makes this case.
Thanks for reading.. will try to rope you into a more detailed conversation next time i see you.
One way to reduce resistance to the use of co-benefits to justify regulatory actions is to take the next step moving beyond Kaldor-Hick efficiency toward Pareto efficiency. The analysis should include the full list of gainers and losers, and financial mechanisms should be designed and implemented to transfer funds from those gainers to losers. For example, a broad income or wealth surcharge could be implemented to compensate power plant owners for premature retirement and loss of capital. This would require a concession by those making gains that those who are losing financially have a “property right” in the their previous use of those environmental attributes before it was found that it was more beneficial to retire their usage. Traditional BCA has tried to dance around this compensation issue, but that failure on the part of economists has led in part to the political logjam that we now face. Time for economists to be more considerate of the distributional impacts of policy proposals.
I always appreciate your blog. How would you apply a cost-benefit approach to PM issues resulting from wildfires in California? (Just now quite obvious and far more grievous than the examples you cite here). Forest derived PM was consciously omitted from the state’s Short Lived Climate Pollution plan, for example.