# 12 Dr. Joseph Shapiro – Is Air Pollution Regulation in the U.S. too Lenient? Evidence from 40 Pollution Offset Markets
Is U.S. air pollution policy still too lenient – even after decades of regulation?
In this episode, I’m joined by Dr. Joseph Shapiro (UC Berkeley, NBER, Energy Institute at Haas) to discuss his recent research using 40 pollution offset markets under the U.S. Clean Air Act. By looking at how much firms actually pay for emission reductions, Joe and his co-authors back out marginal abatement cost curves and compare them to the health and welfare benefits of cleaner air.
We talk about:
How pollution offset markets work in practice, and why they exist at all
Using market prices (instead of just engineering models) to estimate the costs of cutting emissions
Comparing those costs with the benefits of avoided mortality and morbidity
Why, in most markets, benefits still exceed costs by a wide margin
The outlier case of Houston’s VOC market – the “Taylor Swift of offsets” – where fracking-driven demand pushed prices through the roof
What all this means for debates on whether air regulations are “too strict” or still too lax
If you’re interested in how economics, regulation, and air pollution intersect – and in actually quantifying the trade-offs – this episode is for you.
For questions, comments or suggestions, you can contact me at arvid.viaene.ce@gmail.com
Transcript
Arvid Viaene: Welcome to another episode with me, your host, Arvid Viaene. A key tool in environmental economics is the cost‑benefit analysis of policy regulations.
The idea is simple. For air pollution, the benefit of regulation is cleaner air; people are healthier and live longer. On the other hand, reducing emissions can be costly for firms, which increases prices and reduces output. In an ideal world, you balance marginal costs imposed by regulations with marginal benefits.
In practice, measuring both can be hard. In a recent American Economic Review paper, today’s guest, Joe Shapiro, tackled this problem in an ingenious way. Today we discuss his research and what it implies for whether air‑pollution policies in the United States are too strong or too lenient.
My guest is Joe Shapiro—associate professor of economics at UC Berkeley; research associate at the National Bureau of Economic Research and the Energy Institute at Haas. He studies climate change, air pollution, clean energy, renewable and exhaustible natural resources, and especially water pollution. His work links to international trade, public finance, and health. He holds a PhD in economics from MIT. Joe, welcome to the podcast.
Joseph Shapiro: Thanks for having me. I’m excited to be here.
Clean Air Act and Offset Markets
Arvid Viaene: I’m very excited about today’s topic. I’ve covered air pollution in China, India, and worldwide, but this is the first one on the United States. There are many interesting things in your paper. To start, could you give us an overview?
Joseph Shapiro: At a high level, we use a non‑standard approach to measure the marginal costs of air‑pollution regulation and compare them against the marginal benefits. We look at a component of the Clean Air Act that’s been largely overlooked in research. If a new polluting factory opens in a city, it must pay an existing factory in the same city emitting the same pollutant to decrease its own emissions.
These are offset markets. They’re useful for research because they include financial transactions where one firm pays another to decrease pollution. Those transactions help us understand marginal cleanup costs across cities, pollutants, and years.
Arvid Viaene: Why has this been overlooked by researchers?
Joseph Shapiro: The Clean Air Act is a defining piece of environmental legislation in the U.S. It’s been around for half a century and is central in environmental economics research. Many landmark papers study different features of the Act.
There are legal and policy studies on offset markets. Researchers for decades have called them among the most important components of the Clean Air Act’s use of market‑based mechanisms. The Act also has large cap‑and‑trade markets like the Acid Rain Program for sulfur pollution from coal or the NOx Budget Trading Program for nitrogen oxides in the eastern U.S. Why haven’t people looked more at these markets? A conjecture is that each market is a little isolated.
You might have a market for nitrogen oxides in San Francisco, another for volatile organic compounds in Houston. There are hundreds of these markets, but each is one region and one pollutant—smaller and more isolated than national markets like the Acid Rain Program.
They’ve also flown under the political radar. Big cap‑and‑trade markets are in the news; they change, get approved or shut down. Offset markets were reformed by the 1990 Clean Air Act amendments, but they’re not a major topic of political controversy and aren’t highlighted in most economics textbooks.
Another reason is data. Obtaining data is far from trivial. Most of the data we analyze have never been summarized or examined by academics, governments, or the private sector. Two states—Texas and California—have public disclosure laws for these transactions. For 14 other states, we purchased data from a broker who manages these transactions.
Lastly, many markets are decentralized: transactions happen firm‑to‑firm rather than through a central operator like an exchange. That makes analysis difficult because you can’t get a complete record in one place—you need to go to firms or intermediaries.
That’s one reason there are fewer papers with transaction data on decentralized markets, in any setting.
Arvid Viaene: It’s a really cool data collection. Economically, these offsets reflect the marginal costs nicely. As Steve Levitt said, once you see a great paper, you wonder why it hadn’t been done before—it’s a great paper.
Joseph Shapiro: Thanks.
How and Why Economists Compare Marginal Costs and Benefits
Arvid Viaene: Could you situate your paper in the literature? Much research focuses on the marginal benefits of air‑pollution regulation, not the costs.
Joseph Shapiro: Let me briefly clarify marginal costs and benefits. If a factory must decrease its emissions, all cleanup costs are costs of air‑pollution regulation. That includes installing control equipment like scrubbers; labor to manage it; materials; and less obvious costs, like decreased productivity of other activities.
Firms might produce lower‑quality products, face delays or uncertainty, change prices—penalizing profits and affecting consumers. All those are marginal costs.
On the benefit side, decreasing emissions improves ambient air quality, which affects health and welfare—reducing premature mortality, hospital admissions and other morbidity; improving child and in‑utero health, which may increase lifetime human capital and earnings.
It also affects firms—pollution can depress agricultural yields, increase capital depreciation, cause acid rain, and damage ecosystems. All those are marginal benefits. Much environmental economics research focuses on benefits. There’s been huge progress in strategies to measure the marginal benefits of air‑pollution regulation.
Researchers isolate causal effects; account for differences in health among affected populations (e.g., mortality displacement or “harvesting”); cast a wider net of benefits; and use strategies like land values, health, and migration.
However, methods to measure the costs of environmental policy have largely not changed in half a century. Researchers usually use engineering estimates—capital and operating costs of control equipment. Those are informative and easy to use, but have limitations: they’re not revealed preference; actual costs may differ; and they ignore other channels like delays, uncertainty, frictions, quality, and prices.
There are a few papers proposing other methods, but they’re not widely adopted and often measure total costs of broad regulations, not marginal costs of small changes. For optimal policy, we need marginal comparisons.
Arvid Viaene: Great explanation.
Joseph Shapiro: I taught this paper in my graduate class and presented it a few times. We also realized during writing that most progress has been in measuring marginal benefits; much less in widely adopted ways to measure marginal costs.
Arvid Viaene: One contribution of your paper is measuring marginal costs via offsets, allowing comparison with marginal benefits to judge whether it is too lenient or not. What did you find?
Joseph Shapiro: First, on benefits: many estimates combine how emissions affect air quality; how air quality affects mortality; and the valuation of mortality (the value of a statistical life). We take several leading models that make those calculations, and several estimates of the value of a statistical life, aggregating them to value mortality and health in dollars. We then compare those marginal benefits to marginal costs derived from offset transactions.
A benefit of our approach is granularity: different estimates of marginal benefits and costs for each pollutant, city, and year. Our broad takeaway: in most markets, the marginal benefits of cleaning up pollution are more than 10 times larger than the marginal costs. That suggests current regulation in these markets is too lenient—allowing more pollution than optimal because benefits remain large at the margin and costs are modest. Not every market shows this.
The clearest exception is Houston’s market for volatile organic compounds. It’s the Taylor Swift of offsets: demand is enormous, supply is limited, and prices become large.
In the late 2000s, hydraulic fracturing made U.S. natural gas prices low. Industries using natural gas expanded. Houston, a center for those industries, saw a boom in petrochemical and ethylene plants. Many large factories wanted to open, paying incumbents to decrease pollution and quickly marching up the marginal abatement cost curve. Low‑hanging fruit was picked; remaining opportunities were more expensive, while marginal benefits didn’t change. But that’s Houston; most markets show marginal benefits an order of magnitude larger than marginal costs.
How Do Models Put Dollar Values on Health
Arvid Viaene: That illustrates the upward‑sloping marginal abatement cost curve nicely. Back to marginal benefits: one thing listeners may not know is how you calculate them. How do you get from improved health outcomes—like mortality—to dollar estimates?
Joseph Shapiro: We use several models to ensure robustness. Each has similar structure. Suppose you emit one ton of a specific pollutant from a specific county—for example, volatile organic compounds from Houston. For non‑specialists: volatile organic compounds (VOCs) are dozens of chemicals that create ground‑level smog. If you spill gasoline and smell nasty fumes, that’s them—benzene, toluene, etc., which increase cancer risks.
The models ask: how does one ton of VOCs from Houston change ambient air pollution, not only in Houston but across Texas and the country, as wind and chemistry disperse pollution? Once we know changes in ambient air quality, we ask how that affects outcomes like mortality, morbidity, and crop yields.
We use concentration‑response functions from epidemiology and economics. We then assign dollar values to endpoints like mortality and morbidity. For mortality—the dominant component in dollar terms—we use the value of a statistical life (people’s willingness to pay to decrease mortality risk).
It’s often estimated from wages across occupations—for example, window washers on the 100th floor receive higher wages than those on the first; coal miners versus office workers.
We sum damages across locations, expressed in dollars, and call those the marginal benefit of abatement (or marginal damages of emissions).
One leading model, AP3, includes all those categories and finds mortality accounts for more than 95% of benefits. Our headline numbers include mortality; sensitivity analyses add other benefits, but they’re much smaller in magnitude.
[~17:21]
Engineering Models vs Offset-Based Costs
Arvid Viaene: Got it. Earlier you said engineering estimates aren’t revealed preference. How do your offset‑based estimates compare? Did you compare them to engineering estimates?
Joseph Shapiro: We did in earlier drafts and can summarize what we did even though it is not in the final version.. There’s a national engineering model called COST used by the EPA and several California agencies. It lists factories, power plants, and other sources; inventories installed pollution controls; and lists other potential technologies with capital and operating costs.
We compared offset prices (our revealed‑preference estimates) for plants engaging in transactions with what the engineering model says they could install and what it would cost. Our takeaway: the engineering model is all over the map, with little relationship to revealed‑preference estimates.
We examined the model and interviewed regulators and engineers. Major limitations include incomplete data on installed and available technologies; practical constraints (e.g., space, plant layout) the model ignores; recommending unrealistically cheap technologies that many plants already had; and outdated inputs (often from the 1980s). The final limitation was that many of the technologies were out of date, as many of the technologies were from the 1980’s, when the model was last updated and the situation on the ground was quite different. In specific settings, an engineer may know more about an individual plant than a national model. But national policymakers can’t send engineers in real time to thousands of plants; they rely on national models.
For illustration: nationally, for nitrogen oxides (NOx), decreasing emissions by one ton yields $40,000–$50,000 in health and welfare benefits. But offset prices show firms receive only $1,000–$4,000 per ton for NOx reductions. The incentive is too weak—society gains tens of thousands, but firms receive only a few thousand. Regulation is more lenient than optimal.
Engineering calculations show the same qualitative pattern (benefits > costs), but ratios fluctuate widely depending on market, technology, and plant.
Why has there been low awareness of the high marginal benefits? [21:50]
Arvid Viaene: You have established that the marginal benefits are still much higher than marginal costs. Why hasn’t there been more awareness that marginal benefits exceed marginal costs? Why not increase stringency?
Joseph Shapiro: Environmental economics is easy to teach partly because some policies sound crazy as written. For example, components of the Clean Air Act make it illegal to consider costs in assigning certain regulations, and courts have upheld versions of that. Some parts consider costs; many others say they can’t be compared. The goal from the laywers perspective wasn’t to equate marginal benefits and costs. They had lots of other goals, such as that it could be enforable and defensible. And in the 1970’s, the goal was just to bring pollution down. There also have not been a lot of comparisons of marginal costs and marginal benefits in the past
We also recognize policy is political. Changing stringency creates winners and losers. In the political economy spectrum, it is important that winners may be diffuse and unaware, especially the effect of decades of air pollution. There is the concept of a statistical life. When someone passes away, there is understandably an enormous amount of sympathy and sadness, but many 100’s of thousands dieing due to air pollution, it is often hard to say whose death was caused by air pollution, because the actual cause of death is not registered as air pollution, it is cardio-vascular disease, heart attack. In contrast, costs for firms are very visible. It’s harder to attribute specific deaths to air pollution, since death certificates list proximate causes. Political economy matters.
[~25:40]
Arvid Viaene: Earlier you offered to share how you and Reed started working on this topic.
Joseph Shapiro: Two sources. Years ago, Reed Walker and I had written papers on the Clean Air Act. He asked if I’d heard of emissions reduction credits—offsets where plants pay each other to emit. I hadn’t. Later I saw a news article about a plant investing in pollution reduction; it mentioned the plant earned a million dollars by selling credits. I looked into it and realized it was these credits.
I reached back out to Reed. Is this the market you talked about? We looked around, found California and Texas data, and it seemed like a gold mine: thousands of bilateral transactions where firms traded pollution for dollars—rare in environmental economics.
Environmental economics studies externalities—missing markets. When a market is created, we can observe prices and quantities, estimate demand and supply, and understand equilibrium. Finding thousands of transactions for air pollution at the heart of the Clean Air Act offered a natural way to learn about marginal costs, alongside existing measures of marginal benefits.
Arvid Viaene: You also interviewed brokers, buyers, regulators. Did you sense offsets were a deal‑breaker for entry or growth, given rising marginal abatement costs?
Joseph Shapiro: Anecdotes are mixed. Decades ago, people believed firms would just install controls and that’s it. Research has contradicted that. Research has shown firms respond on many margins: exit, avoid entry, switch products, use less energy, reduce output, fire workers, invest less.
In this market, we heard anecdotes of multinational firms considering entry, observing offset prices, and changing plans. Some of these plans were to make plants smaller to buy fewer offsets, enter markets with lower prices, or install extra controls. These are optimizing responses; whether they align with policymakers’ intentions is debatable. If decreasing pollution has large benefits relative to costs, a range of responses can be welfare‑improving.
Talking to policymakers, brokers, and Fortune 500 environmental managers is always revealing. I cold‑called every participant I could find in Houston one day. You learn market details and relevance. I always hold my breath when I do this, to see if how I characterized the market holds up. And every time it is reaffirming, but I do learn some details I did not know about.
Surprises and the Challenges for More Regional Regulations
Arvid Viaene: Anything that surprised you?
Joseph Shapiro: The magnitude and systematic differences between marginal benefits and costs were striking. Houston VOCs were exceptionally high, attributable to fracking’s indirect effects. I also hadn’t appreciated the lack of methods to measure marginal costs before writing this. Offset markets offer an appealing setting to learn costs. They have quirks and differ from cap‑and‑trade, but other environmental markets beyond taxes and cap‑and‑trade can also yield insights.
Arvid Viaene: Two final questions. One: in an ideal world, you could vary regulation by region. But regulation is set nationally. Would more geographic tailoring be ideal?
Joseph Shapiro: There are a lot of economics papers that look at environmental policy and say, wow, this is inefficient and suboptimal and poorly designed. Look at the crazy policies we have. That’s a very common type of conclusion. I increasingly feel that being cognizant of the political economy of environmental policy can add insight to those kind of conclusions besides just saying, hey, this policy is inefficient. And recognizing political economy, there’s a trade-off to allowing local policy It does allow for more flexibility that could tailor to local circumstances. It might also make regulatory capture easier. It makes political sway quite relevant. So local elections can sway policy back and forth. So a feature of the Clean Air Act is that it ties the hands of local regulators. If their air is polluted, the Clean Air Act does not give them a choice. They have to regulate more strictly.
And as this paper is trying to show, it’s doing it with a pretty ah rough tool, which is not perfectly tailored to be optimal. On the other hand, it’s certainly a welfare gain relative to doing nothing or weaker policy. And the more flexibility you allow, the more differences you may have in policy across jurisdictions.
In addition, on top of what the federal government is doing, many local governments have a wide swath of air pollution regulation. There’s a paper like from a couple decades ago that just looks at oil refinery, ah regulations of air pollution for oil refineries in Los Angeles.
And this paper counted more than 40 individual local regulations for oil refineries emitting air pollution. So that’s ignoring state and federal policies. It’s ignoring water pollution. It’s ignoring other industries.
So if there are, and Los Angeles is one of the most tightly regulated environmental settings in the country, but it’s not that every place is doing exactly the same thing. It’s that the Clean Air Act has some degree of uniformity and then local governments can add on top of it.
Many papers conclude environmental policy is inefficient or poorly designed. Recognizing political economy adds insight beyond that. Allowing local policy offers flexibility tailored to local conditions but risks regulatory capture and political swings.
A feature of the Clean Air Act is tying local regulators’ hands: if air is polluted, they must regulate more strictly. It’s a rough tool—imperfectly optimal—but better than doing nothing or having weaker policy.
Local governments also add policies. For example, Los Angeles once had more than 40 local regulations for oil refineries’ air pollution—ignoring state and federal policies and other pollutants or industries. The Act has some uniformity; locals add on top.
Arvid Viaene: That helps. As economists we want to optimize, but there’s a trade‑off between flexibility and regulatory capture. Final question: anything else you’d like to add?
Joseph Shapiro: The conversation is ongoing. Every five years, the federal government must evaluate evidence and decide if standards should be tighter. Then comes a Groundhog Day debate about jobs versus environment. The more we understand health and welfare consequences and the costs of control, the easier it is to assess trade‑offs.
This discussion will keep happening because the law requires periodic reevaluation. Other countries have Clean Air Acts with similar features; this isn’t just a U.S. issue.
Arvid Viaene: Sounds good. Thanks so much. Your paper helps us have more informed conversations. Thanks for taking the time.
Joseph Shapiro: Sure—thanks for reaching out and great questions. It’s fun to chat.