Emissions Cheating, Air Pollution, and Health with Hannes Schwandt

Episode Summary

Executive Summary: The episode explores Hannah Schwant and Diane Alexander’s study on the health effects of Volkswagen diesel emissions cheating. Using county-level car registration data as a natural experiment, they find that unexpectedly high real-world pollution from cheat devices increased low birth weight and childhood asthma, suggesting conventional correlational studies may understate car pollution’s harms and that cleaner transport has broad, immediate public-health benefits.

Main Topics: Why economists study health (Priority: 5/5): Schwant explains that health is central to welfare and productivity, and economists add value by addressing endogeneity and distinguishing causation from correlation in health outcomes. The emissions-cheating scandal as a natural experiment (Priority: 5/5): The VW scandal created a large, unintended pollution shock because consumers and regulators were misled about emissions, allowing researchers to study causal health effects without typical avoidance or selection behavior. Identification strategy and measurement challenges (Priority: 5/5): The study uses county-level registration data and within-county changes over time to handle selection, while an instrumental-variable approach helps address severe measurement error in car-pollution data. Health impacts on infants and children (Priority: 5/5): The paper finds lower birth weight, higher low-birth-weight rates, and more severe asthma attacks among children exposed to the extra pollution from cheating diesel vehicles. Why causal estimates exceed correlational ones (Priority: 4/5): Schwant argues that ordinary least squares likely understates true effects because pollution is measured with error and because quasi-experimental variation reveals larger impacts than simple correlations. Broader lessons for pollution policy and future research (Priority: 4/5): The discussion links car pollution to immediate health harms, notes benefits of cleaner vehicles, and calls for better pollution monitoring and more careful conversion of pollution estimates into health effects. Economics and interdisciplinary health research (Priority: 3/5): The episode closes by stressing that economists identify effects using quasi-experimental methods, while biology and medicine are needed to open the black box of mechanisms.

Key Arguments: Economists study health because it is both a key determinant of productivity and a major component of welfare, but health outcomes are highly endogenous and require causal tools to interpret correctly. The Volkswagen emissions scandal is valuable for research because the cars polluted far more than advertised, people did not know this, and pollution therefore spread without the usual behavioral avoidance responses. County-level registration data let the researchers match where the cars were sold and infer where they were driven, creating geographically varied exposure that supports causal inference. The cheating diesels formed a strong placebo comparison with similar gasoline versions of the same cars, helping isolate the pollution effect from consumer selection into vehicle type. For each additional cheating diesel car per 1,000 vehicles, birth weight fell and low birth weight rose, showing measurable infant-health harm from increased car pollution. Childhood asthma attacks rose substantially in response to the pollution shock, consistent with asthma being a sensitive indicator of air-quality exposure. Instrumental-variable estimates are much larger than ordinary correlational estimates, suggesting that standard air-pollution studies may understate true harms because of measurement error. Effects were not limited to disadvantaged groups; they appeared across socioeconomic strata and were sometimes stronger among more advantaged mothers, implying car pollution is a society-wide health threat. The study’s findings support policies that reduce emissions from cars, including cleaner engines, hybrids, and electric vehicles, even when climate benefits are not the only concern. Economists can identify causal effects, but medical and biological fields are needed to understand the mechanisms behind the observed health outcomes.

Data Points: Cars sold in the scandal: ~600,000 - Approximate number of cheating diesel cars sold in the U.S. Relative pollution vs. normal diesel: Up to 40x NOx emissions - Cheating diesel cars emitted far more nitrogen oxides than standard diesel cars. Relative pollution vs. gasoline car: Up to 160x as much - Compared with a standard gasoline car, cheating diesel cars could pollute dramatically more. Baseline pollution shock interpretation: 1 additional cheating car per 1,000 cars ≈ 10% increase in car pollution - The paper’s baseline estimate translates vehicle counts into an implied pollution increase. Birth weight effect: -6.2 grams - Change in birth weight per additional cheating diesel car per 1,000 cars. Low birth weight effect: +2% - Increase in the rate of low birth weight per additional cheating diesel car per 1,000 cars. Estimated low-birth-weight births: Almost 40,000 - Estimated additional low-birth-weight infants attributable to the 600,000 cheating vehicles. Asthma effect: +8% - Increase in asthma attacks among young children from the 10% pollution increase caused by the cheating cars. OLS vs. IV comparison: 8–10x larger - Instrumental-variable estimates exceeded ordinary correlational estimates by roughly eight to ten times. Geographic pattern: County-level registration data - The study uses county registration records to locate the cars and measure exposure.

Pivotal Quotes: "Health is really central for our lives, and we are willing to pay a lot of money to maintain good health." — Hannah Schwant: Explaining why economists study health outcomes and their welfare importance. "These cars polluted much more than other cars, and people didn’t know about it. So they couldn’t avoid the pollution." — Hannah Schwant: Describing why the emissions cheating scandal creates a useful natural experiment. "Our estimates are eight to ten times as large as the corresponding correlational estimate." — Hannah Schwant: Summarizing how causal estimates differ from simple correlations due to measurement and selection issues.

Implications: The episode suggests car pollution is a larger and more immediate health risk than simple correlations imply. Better emissions monitoring, stricter vehicle regulation, and cleaner transportation technologies could deliver broad public-health gains.