How Does Electric Vehicle Adoption Improve Air Quality and Infant Health?

The study asks whether increased adoption of electric vehicles (EVs) improves air quality and infant health in the United States. Using county-level data from 2010–2021 on EV registrations, air pollution (especially nitrogen dioxide), birth outcomes, and child hospital visits, the authors analyze both fixed-effects and instrumental-variable models. They find that a one standard deviation increase in EV adoption—about 12 per 1,000 vehicles—reduces nitrogen dioxide levels by up to 4 percent, very low birth weight by up to 2.6 percent, and asthma-related emergency visits among children under five by about 11 percent, showing clear health benefits from EV use.

The Policy Scientist’s Perspective

This study addresses one of the most consequential policy questions of our time—the measurable health benefits of decarbonizing transportation. The link between vehicle electrification, air quality, and population health extends well beyond environmental regulation to core issues of economic efficiency, energy transition, and human capital formation. This NBER Working Paper uses exceptionally rich U.S. county-level data on EV registrations, pollution, and infant health from 2010–2021. Its dual strategy of two-way fixed effects and instrumental variables (using Alternative Fuel Corridors) provides credible causal inference. The results are both statistically strong and policy-relevant, showing consistent reductions in nitrogen dioxide and improvements in infant outcomes. The dataset’s scope enhances generalizability to other advanced economies with similar energy profiles, making this one of the most substantively significant empirical contributions on environmental health released this year.

Full Citation and Link to Article

Baran, C., Currie, J., Dursun, B., & Tekin, E. (2025). Clean rides, healthy lives: The impact of electric vehicle adoption on air quality and infant health (NBER Working Paper No. 34278). National Bureau of Economic Research. https://doi.org/10.3386/w34278

Extended Summary

Central Research Question

This study investigates whether the adoption of electric vehicles (EVs) in the United States has led to measurable improvements in air quality and infant health. The authors ask a causal question that goes beyond modeling or prediction: has increased EV penetration, through its reduction of tailpipe emissions, tangibly lowered local pollution and improved birth outcomes? By focusing on nitrogen dioxide (NO₂) as a key pollutant linked to vehicular emissions, the study examines how the shift from internal combustion engines to electric vehicles affects two domains of public welfare—environmental quality and population health. The research also extends the inquiry to young children, testing whether reductions in traffic-related pollutants lead to fewer asthma-related emergency department (ED) visits. This question is important not only for environmental and health policy but also for evaluating the broader societal payoffs of transitioning to cleaner transportation technologies.

Previous Literature

The authors situate their study within an established body of research connecting vehicular pollution to health outcomes, particularly among infants and young children. Foundational studies by Currie and Walker (2011) and Knittel et al. (2016) demonstrated that traffic-related air pollution adversely affects birth weight, prematurity, and infant mortality. These papers leveraged quasi-experimental designs—such as toll plaza reforms or traffic and weather instruments—to isolate causal impacts. Simeonova et al. (2021) similarly found that Stockholm’s congestion pricing policy reduced asthma admissions in young children. More recently, Alexander and Schwandt (2022) used the Volkswagen emissions scandal as a natural experiment, finding that fraudulent diesel emissions increased infant mortality and asthma visits, thereby reinforcing the causal chain between vehicular pollutants and child health.

However, most prior research focused on local or single-state interventions rather than nationwide trends. Moreover, much of the literature on the health benefits of EV adoption has been simulation-based rather than empirical. Modeling studies (e.g., Peters et al., 2020; Choma et al., 2021; Schmitt et al., 2024) projected large hypothetical reductions in pollution under varying EV adoption and energy-mix scenarios, but few evaluated realized health effects. An exception is Garcia et al. (2023), who used California zip-code data and random-effects models to link zero-emission vehicle adoption to reductions in NO₂ and asthma visits. Yet these studies were geographically limited and did not employ causal identification strategies.

The current paper builds on these contributions by offering the first nationwide empirical evidence linking EV adoption to improvements in air quality and health. It extends earlier quasi-experimental approaches by using the staggered federal rollout of charging infrastructure—the Alternative Fuel Corridors (AFC) program—as an exogenous source of variation in EV uptake. By combining high-resolution environmental and health data across more than a decade, the study moves beyond projections to quantify realized, causal effects.

Data

The authors assemble a comprehensive dataset spanning 2010–2021 that links EV registrations, air pollution measures, birth records, and emergency department data. The primary measure of EV adoption is drawn from S&P Global’s standardized vehicle registration records at the county level, allowing the calculation of EVs per 1,000 registered vehicles. This measure captures both battery electric and plug-in hybrid vehicles.

Air quality data come from the U.S. Environmental Protection Agency’s Air Quality System (AQS), which provides daily NO₂ and PM2.5 readings from ground monitors. The authors focus on NO₂ because it is most closely tied to vehicle emissions and displays higher spatial sensitivity to changes in traffic composition. For counties without monitoring data, the authors supplement with satellite-derived pollution estimates (van Donkelaar et al., 2021). They also include county-level NOₓ emissions from power plants using EPA Clean Air Markets data to control for upstream pollution related to electricity generation.

Infant health outcomes are drawn from restricted-use U.S. Vital Statistics Natality files, covering all live births nationwide. Key measures include the incidence of very low birth weight (VLBW, <1500 grams) and very premature birth (<32 weeks’ gestation), expressed per 1,000 live births. The dataset also includes maternal age, race, education, smoking status, and other controls.

For children’s health, the authors use State Emergency Department Databases (SEDD) from nine states, containing all outpatient ED visits from 2010–2021. The primary outcome is the quarterly rate of asthma-related ED visits among children aged 0–5. The dataset also includes visits for acute respiratory conditions (a broader category) and injury-related visits (used as a placebo outcome).

Additional controls include weather variables (temperature, precipitation, wind speed, direction), county-level poverty rates, and vehicle miles traveled by state. This integration of administrative, environmental, and health data at monthly or quarterly intervals provides a robust basis for causal analysis with high temporal and spatial resolution.

Methods

The empirical strategy employs two complementary identification approaches: two-way fixed effects (TWFE) and instrumental variables (IV).

The TWFE models exploit within-county variation over time to estimate how changes in EV adoption affect air quality and infant health, controlling for county fixed effects, month-by-year fixed effects, and state-by-year fixed effects. This structure accounts for unobserved local characteristics and broader temporal shocks. The key assumption is parallel trends—i.e., that counties with differing EV adoption trajectories would have experienced similar pollution and health trends absent adoption. Event-study analyses confirm this assumption by showing no pre-treatment differences in outcomes.

To strengthen causal inference, the authors introduce an instrumental variable approach that uses the federally coordinated rollout of Alternative Fuel Corridors as an exogenous shock to EV adoption. The AFC program, launched under the 2015 FAST Act, required states to build charging stations at fixed intervals (within one mile of highways, spaced ≤50 miles apart). Because the corridor locations were determined primarily by transportation connectivity rather than local environmental or socioeconomic factors, their introduction serves as a plausibly exogenous determinant of EV penetration. The instrument is measured as the number of AFC charging stations per 10,000 residents in a county-month.

The IV approach thus isolates variation in EV adoption attributable to infrastructure access, mitigating potential endogeneity from local preferences or policy targeting. The first-stage regressions show strong predictive power, with a Kleibergen–Paap F-statistic of 17.9. Both TWFE and IV estimates are computed for NO₂ outcomes and subsequently for infant and child health outcomes, using population or birth-weighted regressions.

This dual-method approach, though not equivalent to a randomized controlled trial, represents one of the strongest available causal identification strategies for large-scale environmental policy evaluation. It combines the internal validity of a quasi-experimental instrument with the external validity of nationwide administrative data.

Findings/Size Effects

Both the TWFE and IV models yield consistent and statistically robust evidence that increased EV adoption improves air quality and child health outcomes.

For air quality, the TWFE models indicate that a one-unit increase in EVs per 1,000 vehicles reduces the NO₂ Air Quality Index by 0.02 points (about 0.14 percent) and mean NO₂ concentration by 0.019 parts per billion (0.27 percent). A one standard deviation increase in EV adoption (≈12 EVs per 1,000) corresponds to a 1.6–3.2 percent reduction in NO₂ levels. The IV estimates are larger—up to a 4 percent decline in NO₂ for the same increase—implying that effects are strongest where EV adoption is driven by infrastructure rollout. No significant effects are found for PM2.5, consistent with the persistence of non-tailpipe particulates from tire and brake wear.

For infant health, a one standard deviation increase in EV adoption reduces very low birth weight by 0.8 percent under TWFE and by 2.6 percent under IV estimates. Similarly, very premature births decline by 0.8–2.6 percent, with larger effects concentrated in high-pollution counties near AFC highways. These magnitudes are meaningful in population health terms and translate to annual economic benefits between $1.2 and $4.0 billion from reductions in VLBW alone.

For young children, an additional EV per 1,000 vehicles reduces asthma-related ED visits by 0.022 per 1,000 children, or roughly 1.1 percent of the mean rate. The broader category of acute respiratory visits shows a smaller and statistically weaker decline, while injury-related visits remain unchanged—confirming that effects are pollutant-specific rather than driven by general healthcare utilization trends.

Supplementary analyses show similar directional results when restricting the sample to urban counties or using alternative scaling (EVs per capita). Robustness checks excluding COVID-19 years, redefining exposure by battery-only EVs, and controlling for non-AFC charging stations yield consistent results. The IV effects are larger in magnitude, implying that pollution and health benefits are most pronounced in higher-baseline pollution environments.

Conclusion

This paper provides the first national-level causal evidence that EV adoption yields significant public health gains through reductions in vehicular air pollution. Using a comprehensive dataset and a robust quasi-experimental design, the authors demonstrate that the substitution of electric for internal combustion vehicles reduces nitrogen dioxide concentrations, lowers the incidence of very low birth weight and prematurity, and decreases asthma-related hospital visits among young children.

The paper’s methodological rigor—combining fixed-effects and instrumental-variable frameworks—produces credible causal estimates that complement, and in some respects exceed, those of earlier localized studies. The findings generalize to other high-income countries with similar pollution exposure and electricity profiles, although the benefits would depend on the degree of grid decarbonization. The evidence indicates that EVs produce immediate local health benefits independent of their longer-term climate advantages, strengthening the economic rationale for accelerating electrification policies.

While not a randomized trial, the design convincingly addresses selection bias and endogeneity, placing it among the strongest observational studies to date on the health consequences of clean transportation. The large-scale linkage of environmental, demographic, and health data over a decade adds to the credibility and reproducibility of the findings. Overall, this study constitutes an important advance in environmental economics and public health policy, showing that technological shifts in the vehicle fleet can deliver measurable, near-term improvements in human well-being.