Health benefits of US light-duty vehicle electrification: Roles of fleet dynamics, clean electricity, and policy timing

Significance

This study proposes a unique perspective in the assessment of the air quality–related health impacts of light-duty vehicle electrification policies by comparing an ambitious electrification policy to dynamic fleet renewal relying only on newer internal combustion engine vehicles, instead of a snapshot of the fleet composition in a fixed year. The quantification of the monetized health impacts of electrification over time (between 2022 and 2050) and space (county-level resolution) can help inform policy design and the associated health outcomes for different electrification targets; the addition of different electricity mix scenarios highlights the importance of backing fleet electrification with the deployment of clean electricity.

Abstract

We present a dynamic perspective to quantify the air quality–related health impacts of the electrification of light-duty vehicles in the United States between 2022 and 2050. Using a fleet turnover model and future electricity generation mix scenarios, we compare ambitious vehicle electrification to fleet renewal relying on newer internal combustion engine vehicles, without electric vehicles. The model includes vehicle-level pollutant emission factors and a reduced complexity air quality and valuation model and covers direct (tailpipe, brake wear, and tire wear) and indirect (production of electricity and liquid fuels) emissions of NO_x, SO₂, PM_2.5, NH₃, and VOCs, with a breakdown at the county level to identify geographical disparities in the distribution of health impacts. Short-term health benefits are mostly generated by reductions in NO_x emissions from newer gasoline vehicles, while fleet electrification generates further benefits in the long term. The electricity mix plays a crucial role in the success of electrification policies. With continued grid decarbonization, electrification would reduce harmful air quality–related health impacts cumulatively by 84 to 188 billion USD over the study period, compared with fleet renewal without electric vehicles. In contrast, artificially freezing the 2022 grid would make electrification responsible for 32 to 71 billion USD additional health disbenefits compared with fleet renewal. Finally, we show that while fleet electrification achieves most of its benefits over fleet renewal in the long term, delaying the implementation of such policies would sacrifice meaningful cumulative benefits.

This study uses a dynamic fleet turnover model and future electricity generation mix scenarios to investigate the contributions of fleet renewal and fleet electrification to reducing the mortality associated with air pollution from the fleet of light-duty vehicles (LDVs) in the United States between 2022 and 2050.

Air pollution is a leading cause of mortality, accounting for nearly 7 million deaths worldwide in 2019 (1), with the majority (4 million) from ambient particulate matter exposure. In the United States, exposure to vehicle-related fine particulate matter (PM_2.5) was responsible for 19,800 deaths in 2017, down from 27,700 in 2008 (2) thanks to stricter emission standards focused on lowering nitrogen oxides (NO_x) and tailpipe PM_2.5 emissions. Chronic PM_2.5 exposure is also linked to an increased risk of heart attacks, asthma, and decreased lung function (3). Besides primary PM_2.5 (emitted directly by the source), secondary PM_2.5 is formed from chemical reactions involving precursor gases, mainly NO_x, sulfur dioxide (SO₂), volatile organic compounds (VOC), and ammonia (NH₃) (4).

Vehicle electrification policies are being developed globally to target reductions in the emissions of greenhouse gases (GHGs) as part of a multisectoral effort to mitigate climate change. The US government targets a 50% market share for zero-emission passenger vehicles (ZEV—a classification including battery electric vehicles, plug-in hybrid vehicles, and fuel cell electric vehicles) by 2030 (5). California and several other states go one step further by implementing the Advanced Clean Cars II regulations within the frame of the zero-emission vehicle (ZEV) mandate (6), targeting a sales share of 100% for ZEVs by 2035 (7).

Fleet electrification policies have the potential to reduce fleet-level tailpipe emissions and the associated mortality, but they also shift emissions from the vehicles to the power plants, while nonexhaust primary PM_2.5 emissions (tire, brake, and road wear, dust resuspension) remain at the vehicle level. Fleet electrification is likely to increase electricity consumption in the United States, representing between 0.65% (conservative scenario) and 38% (high electrification scenario) of the total electricity consumption by 2050 (8). In 2022, the electricity sector was responsible for 11% of total US NO_x emissions, and 50% of the total SO₂ emissions, with the overwhelming majority of the SO₂ generated by coal power plants (96%) (9).

Past studies have shown potentially large reductions in air pollution-related mortality from replacing internal combustion engine vehicles (ICEVs) with battery electric vehicles (BEVs) (10–23); however, most quantify the impacts of electrification policies against a snapshot of a past or present LDV fleet used as the baseline. Few studies consider a dynamic fleet in which older ICEVs are already being replaced with newer ones, against which BEVs will have to compete. Against this backdrop, projected air quality impacts of fleet electrification (compared to a counterfactual future fleet without BEVs) remain unclear.

High-resolution chemical transport models (CTMs) have been commonly used to assess the impact of BEVs on the emissions and the concentrations of pollutants at a local scale (city, metropolitan area). Studies focusing on densely populated regions such as the Greater Toronto and Hamilton Area, (10, 11) the Greater Houston Area (12), and California (13, 14) highlighted the local health benefits generated by BEVs through reductions in NO_x and PM_2.5 concentrations, resulting in fewer cases of mortality and asthma. Such high-resolution models are critical for evaluating local effects, especially with respect to environmental justice and vulnerable populations; however, they require high computational resources and lack the flexibility to assess numerous scenarios at a larger scale, including dynamic fleet evolution, to efficiently inform policies at the state and national level.

Schnell et al. (15) developed a larger-scale model to calculate the impacts of an immediate replacement of a fraction of ICEVs with BEVs on the concentrations of ozone and PM_2.5 over the United States, including emissions from power plants. Peters et al. (16) coupled the above model with BenMAP (17) to quantify the monetized benefits from changes in concentrations of pollutants and avoided CO₂ emissions following fleet electrification. Their model does not feature a dynamic fleet renewal component to examine the evolution of the health impacts over the years leading to a highly electrified fleet.

Reduced complexity models are an alternative to CTMs, as they feature a lower spatial and temporal resolution but also require less computational power, making them able to compare different policies at a larger scale (18, 19). Gallagher et al. (20) used the COBRA model to assess the spatial distribution of the health benefits of multisectoral decarbonization strategies in the United States and their impacts on environmental justice. Choma et al. (21) used the InMAP model to quantify the monetized benefits of LDV electrification in major US metropolitan areas and compared them to the associated reductions in GHG emissions. Using other reduced-complexity models, AP2 and EASIUR, Tong and Azevedo (22) quantified the damages from emissions of GHG and criteria air pollutants from different engine technologies in each US county, highlighting that no engine technology was able to simultaneously maximize climate and health benefits, and showing the impact of the composition of the electricity mix on the cobenefits provided by BEVs. None of these studies considered a dynamic age-based replacement of older vehicles with newer ICEVs or BEVs to quantify the impacts of policy implementation timing and compare the benefits of BEVs with newer ICEVs.

The only example of a dynamic fleet model being used to quantify the benefits of fleet electrification was developed by Mehlig et al. (23) to quantify the emissions of NO_x and PM_2.5 in an electrifying fleet of LDVs in the UK and compare them with emissions of a fleet without BEVs, relying on newer ICEVs that follow stricter emission standards. They found significant reductions in NO_x in both scenarios, but largely unchanged emissions of PM_2.5. Their work focused on direct emissions from vehicles and did not include emissions from the production of electricity and liquid fuels.

Given the above research gaps, we develop a model to investigate the dynamic evolution of the air quality health cobenefits of different fleet electrification policies at the national level in the United States considering different pathways for the electricity generation mix. We couple the dynamic fleet evolution and fuel consumption components of a fleet-based life cycle assessment (LCA) model (FLAME) (24), with scenarios on the future evolution of the electricity mix, vehicle-level emission factors for NO_x, SO₂, PM_2.5, VOC, and NH₃ (MOVES) (25), and a reduced-complexity air quality model (COBRA) (26) to quantify the cumulative health impacts from chronic exposure to primary and secondary PM_2.5 of fleet electrification scenarios between 2022 and 2050. The model covers the contiguous United States with a spatial resolution at the county level. We consider the direct emissions from the fleet of LDVs and the indirect emissions linked to the production of the electricity and liquid fuels required to power the fleet. Our baseline scenario simulates a fleet with a 100% sales share of new ICEVs to replace older vehicles, reflecting the dynamics of fleet renewal and the impact of the latest emission standards in the absence of electrification. We then consider a set of electrification scenarios under different geographic scopes and grid mixes (Table 1).

Table 1.

Fleet electrification scenarios

Scenario	Fleet electrification level	Electricity mix (Source)
Renewal	No BEVs–only ICEVs	Mid-case (NREL)
AEO	Nationwide low electrification (2021 AEO)	Mid-case (NREL)
Limited ZEV	15 states follow the ZEV policy*	Mid-case high electrification (NREL)
Extended ZEV	All states follow the ZEV policy	Mid-case high electrification (NREL)
2022 mix	All states follow the ZEV policy	2022 mix
Low renewable energy cost	All states follow the ZEV policy	Low renewable energy cost (NREL)
No inflation reduction act	All states follow the ZEV policy	No inflation reduction act (AEO 2023)
High coal share	All states follow the ZEV policy	Retirement constraints; Reference Electrification (NREL-EFS)

^*States following the ZEV policy: California, New York, Massachusetts, Vermont, Maine, Pennsylvania, Connecticut, Rhode Island, Washington, Oregon, New Jersey, Maryland, Delaware, Colorado, and Minnesota.

The detailed compositions of the fleet and of the electricity mixes are provided in the Methods section. ZEV policy refers to achieving 100% BEV share of new vehicle sales by 2035.

We compare the health impacts of electrification with a fleet composed only of ICEVs by investigating the dynamics of the yearly emissions and associated mortality in the baseline scenario and in an ambitious electrification scenario. Then, we quantify the cumulative health impacts over the 2022–2050 period and their geographical distribution in different electrification scenarios and highlight the influence of the electricity mix. Next, we compare the lifetime monetized benefits of BEVs (in 2023 USD) as a function of their year of introduction into the fleet, and finally, we analyze the consequences of delaying the implementation of electrification policies and the deployment of clean electricity.

Results and Discussion

Newer Internal Combustion Engine Vehicles Drive Early Reductions in PM_2.5 Attributable Mortality; Vehicle Electrification Is Needed in the Long Run.

The pollutants emitted by the LDV fleet (light-duty passenger cars and trucks), including the production of electricity for BEVs and liquid fuels for ICEVs, are responsible for significant health damages in the form of increased PM_2.5 attributable mortality, also referred to as mortality (Fig. 1A). We also expressed them in the form of monetized health impacts (Fig. 1B), representing more than 95% of total health impacts from air pollution (other impacts are nonlethal pulmonary and cardiac diseases, asthma, not included here), with an estimated 3,000 deaths per year and 34 billion (bn.) 2023 USD in annual health damages in the first year of the simulation (2022). These figures do not include indirect impacts caused by the emission of GHGs (social cost of carbon). The adverse health impacts of LDVs have been declining in the past years and will continue to decline in the near future, as a result of stricter emission standards and the replacement of old ICEVs.

Fig. 1.

Cumulative health impacts of the emissions of pollutants from the fleet of LDVs in the “Renewal” and the “Extended ZEV” scenarios between 2022 and 2035, showing the yearly and cumulative mortality (A) and the present value of monetized health impacts (B). The Extended ZEV scenario assumes a 100% share of BEVs in new sales by 2035, with a linear increase of the share between 2022 and 2050. The fleet composition (total fleet size, relative share of light passenger trucks and passenger cars) follows the predictions from the 2021 Annual Energy Outlook (27), and the electricity mix follows scenarios from the National Renewable Energy Laboratory (28) (Mid-case projection for the Renewal scenario and Mid-case High-electrification projection for the Extended ZEV scenario). The Renewal scenario has only ICEVs. In both scenarios, vehicles removed from the fleet (following age-based scrap rates) have been replaced with newer vehicles: only ICEVs in the Renewal scenario and a variable share of BEVs in the Extended ZEV scenario. The monetized health benefits, calculated with COBRA (26) and given in 2023 USD, have been calculated with a 2% discount rate, taking 2023 as the starting year. In both figures, the solid lines show the yearly impacts (Left axis) and the dotted lines the cumulative impacts (Right axis).

The mortality was calculated as the average between a lower and a higher estimate based on two nonthreshold log-linear concentration–response functions (CRFs) (29, 30) and representing the higher and lower bounds of the health outcomes following changes in PM_2.5 concentrations. Infant mortality, based on one CRF (31), was added to the upper and lower bounds. A discussion on the choice of the CRFs is available in SI Appendix, section 3.1. Qualitative conclusions throughout the paper are robust to the choice of the CRF. We chose to show both values in Fig. 2 to highlight the uncertainty in the CRF in our scenarios; to improve readability, we use their average values in the other figures.

Fig. 2.

Cumulative monetized health impacts over the 2022–2050 period considering a 2% discount rate, calculated for the scenarios presented in Table 1. Negative values indicate a decrease in mortality compared to the reference scenario. The scenarios show the consequences of an increased level of electrification in (A) and the impact of the electricity mix in (B). The two values indicated for each scenario correspond to the high and low estimation of the health impacts calculated by COBRA (26) and represent the uncertainty in the concentration–response functions for PM_2.5 attributable mortality. All health impacts are relative to the Renewal scenario, accounting for correlated uncertainties (i.e., comparing the high and low from each scenario, respectively, to the high and low in the renewal baseline). Note that the hypothetical scenario in which the 2022 fleet persists as-is shows larger health impacts beyond the boundaries of the graph, indicating that all scenarios result in large improvements over the current fleet.

We developed two scenarios to assess the ability of an ambitious electrification policy (Extended ZEV scenario, assuming a 100% share of BEVs in new sales nationwide by 2035) to provide higher reductions in mortality than fleet renewal without BEVs and relying on newer ICEVs (Renewal scenario). The baseline for the fleet composition is outlined in the figure caption and additional details on the fleet and the electricity mix are available in the Materials and Methods and SI Appendix, sections 1.3 and 1.4.

The yearly evolution of the health damages is similar in both scenarios until 2028, with 2,210 deaths per year in the Renewal scenario, and 2,130 deaths in the Extended ZEV scenario, respectively, 23.2 and 22.3 bn. USD in yearly health damages, before the curves start to diverge. The yearly mortality reaches a minimum in the Renewal scenario once most of the older vehicles have been replaced with lower-emitting ICEVs, before rising again in the later years (after 2039) because of the increasing fleet size and the increasing market shares of passenger trucks (these two parameters follow the predictions from the 2021 Annual Energy Outlook (27) and are unchanged in our scenarios, and thus highlights the risk associated with the consumer shift to a larger and heavier vehicle class). The yearly mortality continues to decrease in the Extended ZEV scenario as the share of BEVs increases and the electric grid becomes cleaner before slowing down when most of the ICEVs have been replaced with BEVs, and tailpipe emissions become a negligible contributor to the fleet’s emissions, dominated by nonexhaust emissions from vehicles and emissions from power plants. The monetized yearly health disbenefits decrease in both scenarios (Fig. 1B). They do not follow the mortality curve (Fig. 1A) in the later years, as a consequence of the discount rate (2% per year), which is especially acute for the simulation years furthest into the future. Choosing 2023 as the reference year allows us to collapse the string of future health impacts into a single value in present-day terms. This is relevant both as a digestible summary of future impacts and as a comparison against which to benchmark mitigation policy costs. This is a conservative approach; to the extent that policy costs may occur in the future (e.g., annual BEV incentives), future health impacts may not require full discounting to 2023, and so could be comparatively larger than our results imply.

In 2050, the yearly mortality decreases to 1,800 in the Renewal scenario and to 800 in the Extended ZEV scenario; the yearly monetized health damages reach, respectively, 13.8 bn. USD, and 6.1 bn. USD. The cumulative mortality in the Renewal scenario reaches 55,900 deaths between 2022 and 2050 and 42,500 deaths in the Extended ZEV scenario. The cumulative monetized mortality reaches 531 bn. USD and 415 bn. USD, respectively.

Reductions in NO_x emissions (by 82% in the Renewal scenario between 2022 and 2050 and 93% in the Extended ZEV scenario) are the major contributor to the decreasing mortality in both scenarios in the early years; however, they become less significant as older vehicles leave the fleet and are replaced with newer ICEVs (the oldest vehicles in the 2022 fleet have an emission factor 100 times higher than the newest ones, while this ratio falls to 4 in the 2039 fleet and to 2 in 2050, SI Appendix, Fig. S48) and BEVs. Detailed data on the emissions by sector are available in SI Appendix, section 2. Primary PM_2.5 emissions decline in the Extended ZEV scenario (by 42% between 2022 and 2050), while they remain stable in the Renewal scenario (2% decrease). The analysis by sector shows that electricity production becomes responsible for most of the LDVs-related NO_x (60%) and SO₂ (78%) emissions in 2050 in the Extended ZEV scenario, while the fleet is directly responsible for the largest share of primary PM_2.5 emissions (72%). In the Renewal scenario, the fleet is still responsible for the majority of NO_x emissions but decreasing tailpipe emission factors increase the relative weight of the production of liquid fuels (52% of total NO_x emissions for the fleet vs 48% for the production of liquid fuels), which is also responsible for the majority of SO₂ emissions (80%). The fleet generates the majority of primary PM_2.5 emissions in the Renewal scenario (74%). The share of brake and tire wear in the primary PM_2.5 emissions increases in the Extended ZEV scenario, from 36% in 2022 to 67% in 2050; despite the substantial increase in relative share, the absolute increase is only 7.6%.

These results show that the benefits of fleet electrification are relatively small compared with fleet renewal in the short term, as most of the improvements result from removing old vehicles from the fleet. As the share of BEVs increases and the grid becomes cleaner, electrification brings notable long-term health benefits compared to renewal. The reduction of adverse health effects reaches a plateau in 2050 once most of the ICEVs are replaced, and future efforts need to target noncombustion emissions, cleaner electricity, and reductions in car usage.

Successful Scale-up of Clean Electricity Is Essential for Enhanced Health Benefits of Electrification Over Fleet Renewal.

We use the scenarios described in Table 1 to investigate the benefits of different fleet electrification policies, and the role of the electricity generation mix. Having shown the evolution of mortality in the Renewal and the Extended ZEV scenarios and seeking to maximize the contribution of fleet electrification in reducing health damages, we adopt the Renewal scenario as our baseline against which electrification scenarios must compete. When referring to electricity sources, we consider low-emitting sources of electricity as sources that emit lower levels of NO_x, SO₂, PM_2.5, NH_3, and VOCs during the generation phase (nuclear, and renewables such as wind, hydropower, and solar; excluding biomass).

The cumulative monetized health impacts in each scenario are shown in Fig. 2, taking the benefits in the Renewal scenario as the baseline. The different electrification policies are compared in Fig. 2A. In addition to the Extended ZEV scenario, the “AEO” scenario represents the predictions from the Energy Information Administration’s 2021 Annual Energy Outlook (27) and features a limited share of BEVs (reaching 6% of the total stock by 2050 and 10% of new sales). In the “Limited ZEV” scenario, some states follow the ZEV policy and others follow the AEO scenario. The three scenarios shown in Fig. 2A follow the Mid-case (AEO scenario) and the Mid-case High electrification (Limited ZEV and Extended ZEV) predictions from the NREL (28). Additional fleet composition scenarios are available in SI Appendix, section 3.3, investigating the impact of the Inflation Reduction Act on the electrification rate.

Under the Mid-case grid, the AEO scenario provides few benefits over the Renewal scenario (reducing the negative health impacts by 3 to 7 bn. USD). The Limited ZEV policy significantly reduces the negative impacts (by 31 to 70 bn. USD) and extending the electrification policy to the entire country further reduces them (by 71 to 160 bn. USD).

The health benefits are highly dependent on the evolution of the electricity mix, as shown in Fig. 2B, based on the Extended ZEV fleet composition scenario. The scenarios were derived from projections of NREL (28), and the Energy Information Administration (32), along with an artificial 2022 mix scenario that assumes that the electricity mix in 2022 will be unchanged until 2050 (details are given in SI Appendix, Fig. S7). The emission factors for each fuel in each eGRID region were kept at their 2021 values throughout 2050 in all our scenarios. In addition to cases focused on average electricity mixes, we approximate the consequential long-run marginal electricity mix associated with electrification by developing a Delta scenario corresponding to the power sources used to generate the additional electricity demand in the Mid-case High electrification compared to the Mid-case scenario. Alternative electricity mix projections are presented in SI Appendix, Section 3.4, including projections derived from the 2023 Annual Energy Outlook (32).

Keeping the current (2022) grid mix and the current power plants’ emission factors results in electrification scenarios yielding worse results than renewing the fleet without BEVs, with relative health damages increasing by 32 to 71 bn. USD compared to the Renewal scenario. The 2022 mix scenario generates higher emissions from electricity production than all projections from NREL (28) and AEO (32). It is a pessimistic estimate of the impacts of electricity production in the ambitious electrification scenario that is included primarily to provide a simple baseline against current conditions. For additional realism, we include two high-emissions projections: 1) the High coal share, based on a scenario developed by NREL in the Electrification Futures Study which assumes that coal and nuclear production capabilities are maintained for a longer time than in the other NREL scenarios (33), and 2) the No Inflation Reduction Act (AEO) projection, which excludes tax credits for renewable electricity, and other incentives to develop low-emitting sources, introduced in 2022 by the Inflation Reduction Act (34). In those two scenarios, fleet electrification generates health impacts closer to the renewal scenario, with a 17 bn. to 38 bn. USD increase in damages using the High coal share projection, and an 18 bn. to 40 bn. USD decrease in damages in the No Inflation Reduction Act projection. Similar results are found for grid mix from the reference cases in past AEO reports (2019–2022), with cumulative benefits as high as 29 bn. or disbenefits as high as 52 bn. relative to fleet renewal (SI Appendix, Fig. S37).

The potential of BEVs to reduce health damages is fully realized with a high share of low-emitting electricity (reduction of the health damages by 84 to 188 bn. USD in the Low renewable energy cost projection). The influence of electricity production is highly dependent on the share of coal in the electricity mix, as it generates high amounts of SO₂, which drives substantial health disbenefits. Damages decrease by 51 to 118 bn. USD in the Delta scenario, which is similar to the AEO 2023 reference case (SI Appendix, Fig. S37). The additional electricity demand in the Delta scenario is met with a higher share of coal (12% over the 2023–2050 period compared to 6% in the Mid-case projection) despite an increase in renewable electricity.

For comparison, keeping the fleet’s emissions at the 2022 level until 2050 would generate significant health damages (198 to 446 bn. USD) compared to the Renewal scenario. Removing all LDVs would generate 324 to 735 bn. USD in benefits. All scenarios are preferable to the present fleet and add critical context for studies that tend to compare BEVs to present-day emission levels. While electrification is preferred over the current fleet, it can only outperform fleet renewal under scenarios where the grid continues to shift toward low-emission technologies.

All Counties Will Experience Overall Average Positive Health Impacts only if BEVs are Powered by Low-emitting Electricity Sources.

The health impacts shown in Fig. 2 were calculated for the entire United States. However, regional differences in population density and local electricity mix affect their spatial distribution. The data in Fig. 3 show the geographical distribution of the cumulative per-capita average health impacts by county, taking the Renewal scenario as the baseline (impacts = 0). The spatial resolution of our model does not go beyond the county level and thus does not reflect disparities within counties. Communities in a county with overall positive impacts from electrification may still experience negative impacts at a local scale (e.g., next to coal and natural gas power plants) while counties with overall negative impacts may still include subregions that benefit from electrification (e.g., near major roads, in highly urbanized areas); vehicle electrification and the associated changes in electricity demand have exposure inequalities implications (35–37) which are not evaluated in this work. The values given in this section therefore correspond to average health impacts.

Fig. 3.

Geographical distribution of the cumulative (2022–2050) county-average per-capita health benefits in the Limited ZEV scenario in (A) with the borders of the states applying the ZEV policy highlighted in red—and the Extended ZEV scenario in (B). The Renewal scenario uses the Mid-case electricity mix scenario; the Limited ZEV and Extended ZEV scenarios use the Mid-case High electrification electricity mix projections. The Extended ZEV scenario is coupled with the Low renewable energy cost mix in (C), with the Delta projection in (D), with the No Inflation Reduction Act in (E), and with the High coal share in (F). The impacts are shown on a per-capita basis, calculated from the differences between the monetized health benefits in the selected scenario and the baseline (Renewal) scenario, divided by each county’s population (adjusted for projected population growth). Positive values indicate average health benefits in each county (i.e., lower mortality) compared to the renewal scenario.

The impacts of the Limited ZEV policy with the Mid-case High electrification electricity mix are shown in Fig. 3A and the states following the ZEV policy have their borders highlighted in red. On average, states that follow the ZEV policy have per-capita benefits of 203 USD over the baseline (Renewal scenario), while nonparticipating states have limited benefits averaging at 73 USD per capita, mostly through a reduction in emissions from the oil and gas industry following the lower demand for liquid fuels. Negative impacts arise locally in some states, resulting from increases in emissions from electricity production: In 0.5% of US counties (accounting for 0.5% of the population), electrification provides fewer benefits than renewal; these counties are all located in states that do not follow the ZEV policy. The remaining 99.5% of counties all have greater overall benefits associated with electrification than renewal.

When the ZEV program (with the Mid-case High electrification electricity mix) is extended to all states in the contiguous United States (Fig. 3B), the average per-capita benefits reach 228 USD, and electrification is better than renewal in 99.7% of the counties (accounting for 99.8% of the US population). The benefits of the electrification scenario are distributed over the entire country. Fleet electrification generates higher average per-capita benefits in densely populated areas (410 USD in New York City, 546 USD in Chicago, and 677 USD in Los Angeles) than in rural areas (25 USD in North Dakota, 108 USD in Kentucky, and 100 USD in Alabama).

The Low renewable energy cost scenario combined with the Extended ZEV policy is shown in Fig. 3C. The average per-capita benefits increase to 268 USD, and electrification outperforms renewal in all counties. Per-capita benefits increase in both urban (448 USD in New York City, 588 USD in Chicago, and 752 USD in Los Angeles) and rural areas (29 USD in North Dakota, 155 USD in Kentucky, and 153 USD in Alabama).

However, with the High coal share projection (Fig. 3F), the average per-capita impacts of fleet electrification are negative and drop to −54 USD, and electrification is worse than renewal in 81% of the counties (accounting for 62% of the population). All areas are not impacted the same way, as New York City and Chicago still experience per-capita benefits from electrification (270 USD and 163 USD, respectively) while Los Angeles (−385 USD) experiences disbenefits. Among the rural areas, the impacts turn negative and reach −9 USD in North Dakota, −275 USD in Kentucky, and −306 USD in Alabama. The details for the other scenarios are given in SI Appendix, Table S10.

The comparison between electrification and fleet renewal shows that geographical disparities could arise across the country, especially between urban and rural areas, without a strong push in favor of low-emission electricity sources. The baseline for the calculation of the health impacts plays a crucial role in estimating the geographical disparities: Compared to the 2022 emissions level, LDV-associated air quality health impacts are overwhelmingly positive (more information in SI Appendix, section 2.5).

BEVs Generate Lifetime Benefits Compared to Newer ICEVs, Even for Vehicles Purchased in the Early Years.

In Fig. 1, we showed that the air quality health impacts were similar in the Renewal and the Extended ZEV scenario in early years (before 2028). In the present section, we quantify the lifetime impacts of a BEV by model year (which is equivalent, in our model, to the year a vehicle enters the fleet) and compare it to an ICEV from the same year (Fig. 4). Even if the cumulative benefits of the Extended ZEV scenario over the Renewal scenario are marginal before 2030 with the Mid-case High electrification electricity mix, a BEV bought in 2023 generates on average 410 USD health benefits compared to an ICEV from the same year (435 USD with the Low Renewable Energy Cost mix, but generates 180 USD higher disbenefits with the High coal share projection). The breakdown by county highlights significant geographical differences (values given for the Mid-case High electrification electric grid, other scenarios are available in SI Appendix, section 5): The benefits per BEV are notably higher in some urban areas, with however significant geographical variations (3,240 USD in New York City, 645 USD in Los Angeles, 1,140 USD in Chicago) than in rural areas (115 USD in North Dakota, −10 USD in Kentucky and −8 USD in Alabama). These benefits increase over the years, as new BEVs are powered by cleaner electricity: The US average lifetime benefits per BEV reach 630 USD for EVs bought in 2050 (695 USD with the Low Renewable Energy Cost mix, but decreased with the 2022 mix to –380 USD, mostly as a result of ICEVs getting cleaner over time. The value increases in urban areas (3,560 USD in New York City, 1,530 USD in Los Angeles, and 1,440 USD in Chicago) and in rural areas (179 USD in North Dakota, 365 USD in Kentucky, and 180 USD in Alabama). Considering the first year only, BEVs introduced in 2023 generate lower health benefits than ICEVs in all grid mix scenarios (−14 USD in the Mid-case High electrification and Low renewable energy cost projections, −28 USD with the High coal share projection).

Fig. 4.

Lifetime health impacts per BEV by model year compared to an ICEV from the same year (reference). The benefits were calculated by comparing the monetized health impacts in the Renewal scenario to the impacts in a scenario where all new ICEVs in a given year were replaced with BEVs. A 2% discount was applied starting in the model year of the vehicle. For vehicles in the fleet after 2050, we estimated the benefits in the missing years keeping all parameters (grid mix, population distribution) at their values in 2050. Mid-case HE refers to the Mid-case High electrification projection. Negative values indicate a decrease in mortality compared to ICEVs of the same year.

The first-year health outcomes for all model years are given in SI Appendix, Table S19 and further illustrate why a single-year snapshot is not sufficient to capture the lifetime impacts of each vehicle.The lifetime health impacts per BEV are dependent on the electricity mix: With the High coal share projection and the 2022 mix, BEVs generate negative lifetime health impacts for each model year. They are positive in the other scenarios, despite being negative in the first year of the vehicle’s life (until 2024 in the Mid-case High electrification and the Low Renewable Energy Cost scenarios, 2028 in the Delta scenario, and 2033 in the No Inflation Reduction Act projection).

The reference year for the application of the discount rate corresponds to the model year of the vehicle, to provide an estimation of the benefits consistent with the purchase-year government incentives to subsidize the purchase of BEVs. Incentives to purchase BEVs bring health benefits over renewal even powered by a high share of high-emitting electricity sources in the early years, under the condition that the grid becomes significantly cleaner during the vehicles’ lifetime.

Delaying the Implementation of Electrification Policies and Clean Electricity can Significantly Erode the Cumulative Health Benefits.

The impacts of delays in reaching the targeted sales share in the Extended ZEV scenario with the Mid-case High electrification mix are shown in Fig. 5A. A delay of 1 y (BEVs sales target of 100% reached in 2036 instead of 2035) reduces the expected cumulative health benefits of fleet electrification over the 2022–2050 period by 7.9 bn. USD compared to the scenario without delay, while a 15-year delay reduces the health benefits by 75.5 bn. USD. Postponing the deployment of low-emitting sources of electricity, shown in Fig. 5B, initially has a smaller impact on the benefits than a similar delay on policy electrification, with cumulative disbenefits between 3 bn. USD (1 y) and 18.5 bn. USD (5 y), but larger delays (>5 y) appear to be much more damaging, reaching 86 bn. USD in losses (15 y) and leading to higher health damages than in the Renewal scenario until 2045. To illustrate the impacts of delaying the deployment of clean electricity, we froze the Mid-case High electrification projection at the 2022 composition for the number of years given in Fig. 5B. This is an artificial scenario that illustrates the impact of delaying changes to the grid itself rather than delaying policy implementation, which would still result in some interim grid changes (e.g., see the No Inflation Reduction Act projection from the AEO). Nevertheless, emissions from the 2022 grid closely track the High coal share scenario for the first 15 y, and so these results represent a plausible high-emission scenario for delayed deployment of low-emission electricity sources. These simplified clean electricity delay scenarios can then be compared with the more realistic projections in Fig. 5C, which shows the yearly evolution of the cumulative health impacts with the electric mix projections from the NREL (28) and the AEO (32). The higher emission scenarios (e.g., High coal share and No Inflation Reduction Act) provide a policy-driven approach to visualize implicit delays in the shift to clean electricity. Similar to panel b, panel c shows relatively minimal difference between scenarios for the first few years owing to both low initial BEV deployment and limited time for the grid scenarios to diverge. Nevertheless, the substantial divergence in later years (leading to a wider spread than in panels a or b) again reinforces the need for near and medium-term efforts to ensure the country follows one of the lower emission grid pathways.

Fig. 5.

Impact of delays in the implementation of the electrification policy (A) and in the deployment of clean electricity sources (B) on the cumulative health impacts (2023 net present value) in the Extended ZEV scenario with the Mid-case High electrification projection. The yearly evolution of health impacts in each scenario is shown in (C). The reference (impacts = 0) is the Renewal scenario. Negative values correspond to a decrease in mortality compared to the renewal scenario. Delays in the electrification policy refer to the year the targeted market share of BEVs in new sales is reached, with the sales share reaching 100% of BEVs in 2035 in the reference scenario; delays in the deployment of clean electricity consist of shifting the composition of the electricity mix to future years. The impact of delayed electrification is limited to the associated vehicle charging emissions and does not consider broader implications for electricity consumption in other sectors.

Fleet electrification does not provide immediate benefits compared to fleet renewal, mostly due to the initially low share of BEVs in the fleet and the higher-emitting electricity mix. However, postponing electrification policies reduces the cumulative benefits substantially over the 2022–2050 period. Due to the slow turnover of the fleet, even a short delay in policy implementation can have negative health repercussions for many years to come. Delayed deployment of clean electricity is less critical for the light-duty vehicle sector as long as the share of BEVs in the fleet is low, but postponing the phase-out of coal and other high-emitting electricity sources for more than 5 y generates considerable cumulative damages, larger than a similar delay in fleet electrification.

To maximize health benefits, fleet electrification policies should be deployed as soon as possible and must be supported by clean electricity sources.

Conclusion

Our results indicate that all fleet electrification scenarios provide reductions in adverse health effects compared to the current fleet. However, this is not an appropriate baseline for the evaluation of electrification policies, as the current fleet will change in the near future with the phasing out of older and higher-emitting vehicles; and evaluating electrification policies against such a baseline may lead to an overestimation of their benefits, overshadowing the crucial role played by the electricity generation mix and the nonexhaust emissions.

Using a dynamic fleet model, we were able to compare the evolution of the emissions and the related changes in PM_2.5 attributable mortality in a fleet electrification scenario and in a hypothetical scenario where fleet electrification did not happen. Stricter emission standards for ICEVs significantly reduce emissions of NO_x, generating most of the health benefits before 2030. In the long run, fleet electrification policies will contribute more to the reduction of air quality–related adverse health effects from LDVs than fleet renewal alone as the number of BEVs on the road increases. However, further reductions in emissions will be undermined by the projected increases in vehicle usage, size, and weight. Nonexhaust emissions will become a major source of vehicle-related primary PM_2.5 and fleet electrification is unlikely to reduce them significantly because of the additional weight of batteries.

Air pollutant emissions associated with the LDV fleet will still lead to a projected 800 deaths per year and 6 bn. USD in annual costs by 2050 in the ambitious electrification scenario. Future policies should focus on reducing the weight of BEVs and encourage alternative methods of transportation (public transit, biking) to reduce the need for passenger cars.

As BEVs shift the emissions from the road to the power plants, the composition of the electricity generation mix becomes a critical factor in assessing the health impacts of electrification, with the largest potential contributor of health disbenefits being SO₂ emissions from coal power plants. Extending the ambitious electrification policy set by the ZEV program can help reduce the adverse health impacts of LDVs, but a clean electricity mix is needed to produce the additional power required by BEVs without eliminating the benefits from electrification.

Extending the ZEV policy to all contiguous US states has the potential to generate significant health benefits, with notable geographical disparities, especially between urban and rural areas. Densely populated urban areas would experience the largest health benefits per capita and per vehicle, but the importance of clean electricity is amplified in these areas, where failure to deploy clean electricity sources would overturn expected benefits and generate significant average health costs compared to newer ICEVs. Ultimately, reducing vehicle use in those areas by developing biking and public transit infrastructure would bring even higher health benefits than electrification. Large differences in per-capita health impacts of LDV electrification between urban and rural areas underline the importance of considering environmental justice in the design of electrification policies.

Even if the early benefits (before 2030) from BEVs appear to be limited compared to newer ICEVs, BEVs bought in 2023 already generate lifetime health benefits exceeding those from new ICEVs, provided that the electricity mix follows any but the most pessimistic grid scenarios (2022 mix or high coal share). Due to the inertia of fleet renewal, it is important to implement fleet electrification policies as soon as possible: Delaying them significantly reduces the cumulative health benefits. Similarly, failing to phase out high-emitting sources and deploy clean electricity sources will undermine the expected benefits of fleet electrification.

Finally, even though GHGs are outside of the scope of the study, their reduction is a major driver of vehicle electrification policies. A comparison of the cumulative GHG emissions in each scenario is shown in SI Appendix, section 4 and indicates that monetized GHG benefits from electrification may surpass air quality health benefits by an order of magnitude—though the local and near-term nature of air quality concerns continue to warrant special attention. Results further indicate that even the Extended ZEV scenario based on the 2022 electric mix would generate considerable cumulative GHG emissions reductions (25%) compared to the Renewal scenario, despite the associated air quality health disbenefits; note that per-vehicle GHG reductions are more substantial than the 25% calculated at the fleet-level, which is partly limited by fleet turnover and the speed of BEV penetration.

We developed metrics to highlight the multifaceted aspects of fleet electrification. We used cumulative and yearly emissions to illustrate the role of timing and emphasize the need for near-term action, even in regions where electrification benefits are modest in the early years. Additionally, we demonstrated that snapshots of the fleet are insufficient to quantify the impacts of electrification in the context of a dynamically changing fleet. The strong interactions between vehicle electrification and electricity production emphasize the need for near-term efforts to promote clean electricity, ensuring that everyone will benefit from policies such as ZEV mandates. Quantifying the lifetime air quality impacts per BEV compared to ICEVs from the same year helps inform cost–benefit analyses of various policies, including ZEV mandates, stricter emission standards, and the development of renewable sources of electricity. Shifting the focus of a discussion usually centered on GHG emissions to spotlight air quality and its direct impacts on human health, can raise awareness of the consequences of electrification among the public and guide individual purchasing decisions.

Limitations.

There are numerous uncertainties associated with the long-term projections described in this study. Where possible, these were addressed via a detailed sensitivity analysis, available in SI Appendix, section 3. Key sensitivities explored include the impact of the choice of the reduced complexity model (i.e., comparing COBRA with other reduced-complexity models—InMAP and EASIUR), background emissions projections, emissions factors for electricity production, and valuation parameters on our scenarios. The sensitivity analysis shows that our conclusions remain valid for alternative fleet composition scenarios and electric grid projections, and are robust to changes in valuation parameters (e.g., discount rate). Improvements in the emission factors from electricity production over time, not accounted for in our main scenarios, improve the health benefits of the scenarios relying largely on high-emitting sources. In particular, a decrease in per power plant emission factors could result in the 2022 mix scenario generating higher benefits than the Renewal scenario (i.e., BEV scenarios become favorable even if the grid mix itself doesn’t change).

Other limitations could not be addressed in this study and are left as areas for future work. The use of a reduced-complexity air quality model is dictated by our scope to investigate various electrification scenarios including different projections for the electricity mix over the contiguous US limits, while keeping the computation time reasonable. This choice limits the spatial and temporal resolution in the calculation of the pollutants’ dispersion and the exposure of the population. The integration of a more complex chemical transport model would allow us to analyze the distribution of health impacts within a county (e.g., in a densely populated urban core, near a highway, high-emitting industries in sparsely populated regions of a county) and further investigate the impacts of fleet electrification and the associated electricity mix projections on exposure inequalities. We did not adapt the source-receptor matrix, linking the emissions to the concentration of PM_2.5, to potential future changes in weather patterns and atmospheric chemistry, impacting the spatial distribution of PM_2.5 and the formation of secondary PM_2.5. The choice of the CRF generates uncertainties around the quantification of the mortality (a discussion in SI Appendix, section 3.1), but our conclusions remain valid with different CRFs, as the conditions under which electrification scenarios are better or worse than the Renewal scenario are still fulfilled.

Our electrification scenarios assume a homogeneous electrification rate across each state, while more realistic scenarios may have different penetration rates for BEVs based on economic and demographic data.

A more realistic model for power generation would take into account the location of new high-emitting power plants as well as the charging patterns of BEVs. Use of average grid emissions is likewise a limitation compared to a consequential assessment of how BEVs would influence future power generation—however, this is partly offset by our use of grid projections that are specific to high rates of electrification, and by the inclusion of a Delta scenario that focuses on the incremental sources used to generate the additional electricity demand in the Mid-case High electrification projection compared to the Mid-case projection. Endogenizing capacity expansion and grid dispatch would add substantial complexity and uncertainty without necessarily improving fidelity. A more accurate estimation of the emissions from refining would integrate economic criteria to select the refineries that would be decommissioned. The emissions related to battery manufacturing and mining activities are not included in this study, their quantification is needed to evaluate the risks associated with the anticipated deployment of mining and manufacturing capabilities in the United States to support electrification policies. A discussion in SI Appendix, section 1.9 highlights the potential damages from the emissions linked to battery manufacturing, and further work is needed to build realistic scenarios and grasp the quickly evolving supply chain.

The fleet model only considers ICEVs and BEVs and other technologies (hybrid, fuel-cell, etc.) are not within the scope of this study due to the lack of vehicle emission factors from MOVES. Uncertainties remain on the emission factors for nonexhaust sources (tire and brake), especially for BEVs. A discussion and a sensitivity analysis are presented in SI Appendix, section 3.5. Road wear and resuspension dust are not included in this study. As the share of BEVs in the fleet increases, brake and tire wear emissions will become the main contributor of primary PM_2.5 and the impact of the changing mix of primary PM_2.5 on the concentration–response functions is unclear. Recent research (38, 39) shows that PM_2.5 related to fuel combustion (black carbon) could have higher toxicity than other sources, which could impact concentration–response functions as noncombustion sources become dominant and generate different health impacts.

Despite these limitations, the main conclusions of the study are clear: Ambitious electrification policies targeting the fleet of LDVs can reduce air quality health impacts compared to fleet renewal with newer ICEVs alone, but only if supported by the deployment of low-emitting sources of electricity.

Materials and Methods

The computational model, named FLAME-AQ in reference to the Fleet Life Cycle Assessment and Material-Flow Estimation (FLAME) (24) model it is built on, developed to calculate the emissions from the fleet of LDVs is summarized in this section. The model includes direct (tailpipe, brake wear, and tire wear) and indirect emissions to produce the electricity and liquid fuels to power the vehicles. Various tools and data sources were coupled to calculate the monetized health impacts linked to changes in emissions. An overview of the model is given in Fig. 6. A comparison of the outputs from FLAME-AQ with other studies is available in SI Appendix, section 6.

Fig. 6.

Overview of the major features included in FLAME-AQ, the model developed to calculate the health impacts of fleet electrification scenarios. Details are available in SI Appendix, section 1. Credits for the icons used in the figure: www.flaticon.com.

The dynamic fleet turnover and the associated fuel consumption calculations are based on the FLAME (24) model; the vehicle-level emission factors were extracted from the Motor Vehicle Emission Simulator (MOVES) (25); and the health impacts of the changes in emissions were calculated with the CO-Benefits Risk Assessment Health Impacts Screening and Mapping Tool (COBRA) (26). COBRA uses emissions of primary PM_2.5, NO_x, SO₂, VOCs, and NH₃ to calculate the health impacts of primary and secondary PM_2.5 at the county level.

Calculation of the Health Impacts.

Mortality from chronic exposure to primary and secondary PM_2.5 was calculated with COBRA (26) (version 4.1), based on emissions of primary PM_2.5, NO_x, SO₂, NH₃, and VOCs. Consequences of acute exposure to pollutants and exposure to ozone are not included. FLAME-AQ generates a baseline and a scenario file for each year between 2023 and 2050, describing the yearly county-level changes in emissions from the fleet of LDVs, the emissions to produce the electricity required by the fleet, and the changes in refining and extraction activity caused by changes in the fleet’s consumption of liquid fuels. The source-receptor matrix built in COBRA translates changes in emissions into changes in PM_2.5 concentrations in each county, and a concentration–response function calculates the mortality associated with changes in the PM_2.5 concentration, as a modifier of the baseline mortality. Projections of the evolution of the baseline mortality, the population, and its age structure by county were obtained from the U.S. Census Bureau (40) and the Center for International Earth Science Information Network (41). Background emissions (sectors not covered in our study, e.g., industrial emissions, construction, agriculture, etc.) followed their historical trends extrapolated until 2050.

The PM_2.5 attributable mortality was extracted from COBRA’s results files. As health impacts are expected to occur over the course of several years beyond the point at which the changes in emissions occur, we ran COBRA with a 2% discount rate according to the latest guidelines from the US government (42), with a value of a statistical life (VSL) for 2023 of 12.1 million USD (calculated from the value recommended by the U.S. EPA (43, 44), i.e., 7.9 million USD in 2008), and we used a VSL elasticity of 0.4 (45) and an income growth rate of 1.3% (46) to project the future evolutions of the VSL until 2050 (details in SI Appendix, section 1.8). All the monetized impacts are given in 2023 USD.

Fleet Turnover and Fuel Consumption.

The historical fleet is based on the 2021 Annual Energy Outlook (AEO) (27), with a breakdown by vehicle size (car or passenger truck), engine technology, and age. The share of ICEVs and BEVs in new sales was modified in the electrification scenarios. The removal of older vehicles follows age-based survival rates included in the FLAME model. States’ allocation factors for each vehicle type (size, engine, and age) extracted from MOVES were used to build the vehicle fleet in each state, with the option to individually modify the share of BEVs in new sales in each state. County allocation factors, also extracted from MOVES, were used to allocate vehicles to counties. They were uniformly applied to all vehicle types.

We consider the following fleet electrification scenarios:

• •Renewal: BEVs were removed from the historical and future fleets and replaced with gasoline ICEVs. The share of BEVs in new sales was kept at 0%.
• •AEO: Followed the predictions from the 2021 AEO until 2050.
• •Limited ZEV: a market share of 100% for BEVs was set in the years 2035 and up, with a linear interpolation between 2022 and 2035. This scenario was applied to the 15 states that participate in the ZEV program (6) (California, New York, Massachusetts, Vermont, Maine, Pennsylvania, Connecticut, Rhode Island, Washington, Oregon, New Jersey, Maryland, Delaware, Colorado, and Minnesota) and the remaining states follow the 2021 AEO.
• •Extended ZEV: the Limited ZEV scenario was applied to all contiguous US states

Emissions from the Fleet.

The emission factors for NO_x, SO₂, PM_2.5 (tailpipe, brake, and tire wear), VOCs, and NH₃ were obtained from MOVES version 3.1.0 (25). The yearly emissions of a reference fleet [the reference scenario of the 2021 AEO (27)] were calculated with MOVES from 2022 to 2050 for passenger cars and light passenger trucks. A postprocessing script was developed to extract the emission factors for each vehicle type for each year between 2022 and 2050 (broken down by state, engine technology, model year between 1970 and 2050, and vehicle size—car or truck). The normalized emission factors reflect differences in yearly driven distance and drive cycles between states.

MOVES attributes similar emissions factors for tire and brake wear PM_2.5 emissions to ICEVs and to BEVs. We further adjusted the PM_2.5 emission factors of BEVs to take into account higher tire wear (+27%) (47) due to the additional mass of batteries, and lower brake wear (−25%) (48) resulting from regenerative braking. There is a high variability in brake wear emission factors for BEVs in the literature, as it is highly dependent on the use of regenerative braking and thus on the driving cycle. A discussion and a sensitivity analysis are available in SI Appendix, Fig. S38. We did not include road wear and road dust resuspension, as these emissions have had significantly less attention in the literature than brake and tire wear.

Emissions from Electricity Production.

As BEVs shift pollutant emissions from the road to power plants, integrating the emissions from electricity production is crucial in estimating the benefits of fleet electrification. The emissions from electricity production rely on the electricity consumption of the fleet, on the electricity mix, and on the emission factors for different sources. The fleet’s electricity demand was calculated by the original FLAME model and was broken down at the county level. The scenarios for the composition of the electric grid were taken from CAMBIUM (28) and the 2023 AEO (32), and power sources were grouped into five categories: coal, natural gas, nuclear, renewables, and others. When referring to electricity sources, we consider low-emitting sources of electricity as sources that emit lower levels of NO_x, SO₂, PM_2.5, NH_3, and VOCs during the generation phase (nuclear, and renewables such as wind, hydropower, and solar). Biomass (representing less than 2% of the predicted electricity generation) was removed from the scenarios, as it generates high amounts of criteria air pollutants despite potentially limiting CO₂ emissions. The electric grid predictions from CAMBIUM are available for each Cambium Generation and Emission Assessment region (GEA) (49) and were broken down by county, assuming that each county has the same electric mix as the corresponding GEA region. Projections from the AEO are available at the national level and broken down at the GEA region level by scaling the electricity production from each source in the GEA regions using data from CAMBIUM. Emissions factors for the generation phase were obtained from eGRID (50) [for NO_x and SO₂, available for each eGRID region), AVERT (51) (VOC and NH₃, attributed to individual power plants using data from the EIA-860 form (52)], and GREET (53) (primary PM_2.5, at the national level). We used the same emission factors until 2050; this parameter is included in our sensitivity analysis (SI Appendix, section 3.7). We calculated the emissions in each GEA region by aggregating the consumption in each county within the GEA region and using the local and national emission factors. We considered that the electricity demand within a GEA region was met exclusively by power plants within that GEA region and that power plants would ramp up proportionally to their output in the initial year to meet the additional demand. We attributed the emissions to each county based on their weight in the total emissions of the corresponding GEA region in the initial year of the simulation. Emissions related to battery manufacturing for BEVs are outside the scope of this study but are bounded via a sensitivity analysis in SI Appendix, section 1.9.

Emissions from Oil Refining.

The evolution of the emissions attributed to the oil and gas industry is based on the yearly fuel consumption of the fleet of LDVs, proportional to its volume in the total demand for refined products. Estimations were conducted at the Petroleum Administration for Defense District (PADD) level with data from the Energy Information Administration (54). Based on flows between PADDs and on the production of crude oil, gasoline, and diesel in each PADD, the changes in the fuel consumption within a PADD were attributed to other PADDs proportionally to the flows of refined products and crude oil entering the PADD. The baseline oil and gas industry emissions were taken from the COBRA emissions inventory for 2023, broken down by the following subcategories: oil and gas production, refining and related industries, petroleum products storage, petroleum products transportation, and service stations. Imports of crude oil and refined products were excluded from the calculation of the emissions. The corresponding emissions were allocated at the county level proportionally to their emissions in the emissions inventory provided by COBRA (for 2023), assuming no changes in the spatial distribution of the emissions until 2050.

Finally, the county-level emissions from the fleet (tailpipe, tire wear, and brake wear), from the electricity needed to power the BEVs, and from the production and distribution of liquid fuels to power the ICEVs, were combined into COBRA’s input files to calculate the yearly air quality health impacts. The cumulative and the discounted monetized impacts were calculated in postprocessing.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The code used to generate the results and all model’s inputs have been deposited in a publicly available repository (https://doi.org/10.5281/zenodo.11507521), as well as results files used to generate the graphs and the discussion (https://doi.org/10.5281/zenodo.12635901) (55, 56).