All podcasts / People I (Mostly) Admire / Summary

171. Measuring Pollution on Parallel Earths

2025-11-22 - source - Read full transcript

Steve Levitt (host)Michael GreenstoneMorgan Levey

Key insights

Fine particulate air pollution (TSP/PM2.5) causes premature mortality at a scale that surprised even the researcher who has studied it for 15-20 years.

Greenstone explains that the smallest particulates released by burning coal and petroleum evade the lungs' natural filtering (cilia) and lodge deep in the body, repeatedly linked to premature death and cardiorespiratory illness. He says the magnitude of the effect he eventually measured still blew him away despite decades in the field.

air-pollution-health-impact

China's Huai River winter-heating policy created a rare, clean natural experiment for measuring pollution's lifetime health cost.

Because China gave free coal-fired winter heating only to areas north of an arbitrary line along the Huai River, and restricted migration across it, people otherwise similar in every way ended up with very different lifetime pollution exposure. Greenstone treats this as the closest real-world approximation to his hypothetical of two parallel earths, one polluted and one clean.

natural-experiments-causal-inference

The pollution gap at the river translated into a 3 to 5.5 year gap in life expectancy.

TSP levels were about 40 percent higher just north of the river (roughly 140 vs. 100 micrograms per cubic meter, against a W.H.O. standard near 10). Greenstone's first paper found a roughly five-year life-expectancy gap; a follow-up with more recent data, after China's pollution levels had fallen, found the gap had narrowed to about three years.

air-pollution-health-impact

Multiple robustness checks (no effect at other latitudes, deaths concentrated in pollution-linked causes) support a causal, not merely correlational, reading of the result.

Greenstone tested for a discontinuity in pollution and life expectancy at every other degree of latitude and found none, only right at the river. The excess deaths on the polluted side were concentrated in cardiorespiratory causes, the category pollution science predicts, rather than spread across unrelated causes like car crashes.

natural-experiments-causal-inference

The Air Quality Life Index (AQLI) was built because the existing Air Quality Index (AQI) fails to communicate long-term stakes.

Greenstone argues AQI's color-coded categories only tell you whether today's air is bad, not what sustained exposure costs you. AQLI instead reports pollution's effect in years of life expectancy under different standards, a unit he says makes the information motivating enough that its annual release 'goes like wildfire' in polluted countries.

research-to-real-world-impact

Cap-and-trade markets beat uniform regulatory mandates because regulators cannot see which polluters can cut emissions cheaply.

Instead of imposing the same abatement requirement on every plant, a cap-and-trade market lets an efficient, low-cost plant sell emissions reductions to an inefficient, high-cost plant, so the same collective pollution target is hit at lower total cost through voluntary trading between firms.

market-based-environmental-policy

A randomized controlled trial of a real emissions market in Surat, India delivered results better than the researchers expected.

Half of 350 plants were randomly assigned into a new sulfur-dioxide trading market designed with the Gujarat Pollution Control Board; the rest stayed under standard command-and-control regulation. Compliance rose from about two-thirds to nearly 100 percent, emissions fell 20 to 30 percent relative to the control group, and total abatement costs for plants fell about 12 percent, an unusually clean win-win result.

market-based-environmental-policy

The market's credibility depended on the regulator actually punishing two large, politically connected firms that violated the rules.

At the end of the first compliance period, 148 of 150 firms held the correct number of permits; the two violators were big and politically connected. Greenstone told the regulator that failing to enforce the pre-agreed penalty would mean no one would ever comply again. The regulator imposed the fine, and every subsequent period saw full compliance.

market-based-environmental-policy

Success in Surat has driven rapid expansion of pollution markets across India, including a statewide sulfur-dioxide market covering large power plants in Maharashtra.

The Gujarat market grew as former control-group plants were folded in, and a second city market launched in Ahmedabad. Greenstone says the world now reads the pilot as proof that cap-and-trade can work even in countries with relatively low state capacity, and inbound requests from other regulators now outpace what his team can handle.

market-based-environmental-policy

Progress on the global warming trajectory over the last decade came mostly from cheap technology, not from strong policy.

Greenstone attributes the shift from a likely 4-5 degree C end-of-century warming path to a likely 3 degree C path mainly to unexpectedly cheap renewables and batteries and to fracking-driven cheap natural gas displacing coal, which has half its carbon content. He calls real policy gains, such as the E.U.'s CO2 cap-and-trade market, real but not 'earth shattering,' especially since 80 percent of future emissions will come from outside the O.E.C.D.

climate-policy-progress

Greenstone rejects hard global temperature caps as what he calls the 'cruelty of climate arithmetic.'

Because most remaining emissions will come from developing countries, a strict cap like 2 degrees C effectively demands those countries cut emissions 80-90 percent below projections, forcing already-poorer populations to pay for expensive energy and accept more impoverished lives. He argues this consequence is rarely reasoned through by people who set such caps, even though few of them intend to impoverish billions of people.

climate-policy-progress

The Tropical Forest Forever Facility's edge-of-forest penalty design is clever but currently underpriced to work.

Levitt explains the fund pays countries $4 per hectare per year for standing forest but deducts $400-800 per hectare destroyed, focusing incentives on marginal edge-of-forest destruction rather than the whole forest. He argues the penalty is still below the roughly $1,000 per hectare that illegal ranchers pay for cleared land, so it may just subsidize Brazil for destruction that would happen anyway, unless the payment gives the government enough of a stake to crack down on the illegal ranchers actually doing the clearing.

research-to-real-world-impact

Media referenced

Shocked - podcast - Greenstone's own new podcast, discussed near the end as a deliberate attempt to get economics research out of the ivory tower after he says he was a poor communicator during his time in the Obama administration.
The Simple Economics of Saving the Amazon Rainforest (Freakonomics Radio, episode 428, 2019) - podcast - Levitt's earlier Freakonomics Radio interview with Greenstone on paying Brazil not to deforest the Amazon; producer Morgan Levey uses it to set up the closing discussion of the Tropical Forest Forever Facility.

Companies

University of Chicago - Greenstone's academic home; he holds the Milton Friedman Chair in the economics department despite a family history of viewing Chicago-school economists with suspicion.
Gujarat Pollution Control Board - The Indian state regulator that partnered with Greenstone's team to run a randomized controlled trial of an emissions-trading market across 350 coal-burning plants in Surat.
Emissions Market Accelerator - Greenstone's initiative to scale pollution-trading markets in developing countries, staffed with implementers rather than researchers, with a stated goal of helping 1 billion people by 2030.

Techniques and frameworks

Geographic regression-discontinuity natural experiment - Greenstone and coauthors exploited China's Huai River winter-heating policy, which gave free coal heat only north of an arbitrary line, to compare otherwise-similar populations exposed to very different lifetime pollution levels.
Cap-and-trade emissions market - Instead of mandating identical pollution-abatement rules for every polluter, the government sets a collective emissions cap and lets firms trade permits, so low-cost abaters sell reductions to high-cost abaters and the total cut is reached more cheaply.
Continuous emissions monitoring systems (CEMS) - Real-time pollution-output sensors India mandated on plants; Greenstone identified this as the missing precondition that made a real cap-and-trade market technically possible.
Air Quality Life Index (AQLI) - Greenstone's tool that converts particulate pollution levels into an intuitive, tangible unit: years of life expectancy gained or lost, built to be more legible and motivating than the color-coded Air Quality Index (AQI).
Tropical Forest Forever Facility (TFFF) marginal-penalty design - A conservation-payment scheme discussed in the closing segment: countries get a small per-hectare annual payment for standing forest but pay a much larger per-hectare penalty for destruction, concentrating the incentive on the marginal, edge-of-forest destruction rather than paying for the whole forest.

Summary

Steve Levitt's guest is University of Chicago economist Michael Greenstone, who has spent his career trying to measure air pollution's true cost to human health and then turn those measurements into working policy. The episode opens with Greenstone's long-standing thought experiment: to really know pollution's lifetime health effect, he would want two parallel earths, one polluted and one clean, with people free to live out full lives and adapt behavior in each. Lacking that, he and coauthors found the next best thing in China's Huai River winter-heating policy, which gave free coal heat only to areas north of an arbitrary boundary and restricted migration across it. The resulting natural experiment showed particulate levels about 40 percent higher just north of the river, and life expectancy gaps of three to five and a half years across two separate studies, with the effect concentrated exactly at the river and in pollution-linked causes of death, evidence Greenstone treats as convincingly causal rather than coincidental.

Greenstone connects this finding to two of his real-world projects. The Air Quality Life Index (AQLI) translates pollution exposure into years of life expectancy, a unit he built specifically because the existing color-coded Air Quality Index tells people whether today's air is bad without conveying what sustained exposure actually costs them. The more ambitious project is the Emissions Market Accelerator, which grew out of his time in the Obama administration working on (ultimately failed) U.S. cap-and-trade legislation. After leaving government, Greenstone pushed India's Gujarat Pollution Control Board to build a real emissions-trading market for coal plants in Surat, structured as a randomized controlled trial: half of 350 plants traded permits under the new market, half stayed under standard regulation. The market outperformed expectations, near-total compliance, a 20-30 percent emissions cut, and 12 percent lower abatement costs, and its credibility hinged on a tense moment where the regulator actually enforced penalties against two large, politically connected non-compliant firms rather than looking away.

That pilot has since scaled rapidly: the Gujarat market absorbed its former control group, a second market launched in Ahmedabad, and Maharashtra is preparing a statewide sulfur-dioxide market covering major power plants, with inbound interest from other regulators now outpacing what Greenstone's team can handle. He frames this shift, from governments viewing market-based pollution regulation as a foreign, developed-world idea to actively requesting it, as evidence that cap-and-trade can work even where state capacity is limited.

On the broader climate picture, Greenstone argues the world's improved end-of-century warming trajectory (from a likely 4-5 degrees C path a decade ago to a likely 3 degrees C path now) owes more to lucky technology shifts, cheap renewables and batteries, and fracking-driven natural gas displacing coal, than to strong policy, though he credits the E.U.'s CO2 cap-and-trade market as a genuine, if not transformative, policy win. He is openly critical of hard global temperature caps, which he calls the "cruelty of climate arithmetic," arguing they effectively force the steepest emissions cuts onto developing countries that will generate most future emissions and can least afford the cost, a consequence he thinks is rarely reasoned through by advocates who propose such caps.

The conversation closes on a more personal note: Greenstone reflects on holding the Milton Friedman Chair despite a family history skeptical of Chicago-school economics, and on why, after a formative early failure communicating with non-economists in the Obama administration, he started his own podcast, Shocked, as another vehicle for getting research out of academia. In the closing listener-question segment, producer Morgan Levey asks Levitt to evaluate the Tropical Forest Forever Facility, a program paying countries a small annual sum per hectare of standing forest but deducting a much larger penalty per hectare destroyed. Levitt argues the penalty price is currently too low relative to what illegal ranchers pay for cleared land to change behavior on its own, though it could still work if it gives the Brazilian government enough financial stake to crack down on the illegal ranching actually driving deforestation.

Notable Quotes

"I was just blown away that the impact of air pollution seems so large. And, if that was news to me, and I'd been studying that for 15 or 20 years, it just felt like, well, that probably would be news to other people as well." - Michael Greenstone

"What I would like to have is two worlds, two earths, and I would like to inject one with high levels of particulates air pollution, and the other, relatively clean air." - Michael Greenstone

"I like to think of using data as, like, the worst boyfriend or girlfriend you've ever had. Like, you know you're going to be disappointed every day. You just don't know how. And so this was an exception to that." - Michael Greenstone

"I'm not that interested in stated preference. I want to know how much you're willing to pay. And to me, that's the ultimate metric of how much you care." - Michael Greenstone

"This would be in many ways the worst possible program - a program that looks like it's really helping, that's really expensive and takes a lot of resources, but doesn't do any good." - Steve Levitt