Economy-wide evaluation of CO₂ and air quality impacts of electrification in the United States

Electrification—substituting electric end-use technologies for fossil-fueled alternatives—is a key element of economy-wide decarbonization alongside efficiency, carbon capture and removal, low-carbon fuels, and demand-side responses. Increased adoption across buildings, industry, and transport is driven by technological change, consumer choice, and CO₂-targeted policies. Beyond CO₂ mitigation, electrification can reduce criteria pollutants from the transportation and electricity sectors, major U.S. sources of NOₓ and SOₓ that form ozone and fine particulate matter (PM). Air quality benefits can be immediate and localized, but the magnitude is uncertain due to the speed and scale of electrification, complex emissions–air quality interactions, and contributions from other sources (e.g., fugitive dust, agriculture, solvents). These factors have large implications for human health and policy design.

The study links EPRI’s U.S. Economy, Greenhouse Gas, and Energy (US-REGEN) model with the Comprehensive Air Quality Model with Extensions (CAMx) to evaluate economy-wide CO₂ and air quality impacts of electrification in the Continental U.S. (CONUS). US-REGEN estimates CO₂ and criteria pollutant emissions for 2015–2050 at five-year intervals with sectoral, fuel, and regional detail, mapping outputs to Source Classification Codes (SCC), county, and NAICS codes. US-REGEN includes: (i) an intertemporal electric sector capacity planning and dispatch model with endogenous investments in generation, storage, transmission, hydrogen production, and CO₂ transport and storage; and (ii) end-use sector models (passenger transport, buildings, other transport, industry) using lagged logit technology choice with structural detail (e.g., building type/vintage, climate zone, vehicle ownership). Electricity load shapes are consistent with 2015 weather; demand shapes vary by sector and technology. Exogenous projections of population, economic activity, and service demand are from U.S. EIA’s Annual Energy Outlook; fossil fuel prices are exogenous; electricity and hydrogen prices are endogenous. Cost and performance assumptions are from EPRI documentation and literature. Emissions include ozone and PM precursors (NOₓ, SOₓ, VOC, CO, NH₃) and primary PM. CAMx v7.0 is applied on a 12-km grid over the lower 48 states nested in a 36-km grid, simulating every hour of calendar year 2016. The same 2016 meteorology is used for all scenarios to attribute air quality changes solely to emissions changes. The 2016 baseline uses EPA’s 2016v1 modeling database (Revised Cross-State Air Pollution Rule Update). Model performance meets standard ozone MDA8 error/bias goals; PM₂.₅ performance is comparable to EPA’s with some winter overestimation. Scenarios (with on-the-books federal and state policies as of June 2021) include: (1) 2035 Limited Electrification (restricted EV adoption, constrained building electrification shares, higher costs for electric tech in industry/heavy-duty transport; flat electricity demand); (2) 2035 High Electrification without Carbon Price (endogenous EV deployment, accelerated heat pump performance, faster stock turnover); (3) 2035 High Electrification with Carbon Price; and (4) 2050 High Electrification with Carbon Price. The carbon price applies economy-wide starting at $50/tCO₂ (2025, 2020 USD) growing 7%/yr to $271/tCO₂ in 2050, layered on existing policies (RPS/CES, offshore wind/storage mandates, California AB32, RGGI, NSPS, ITC/PTC). Air quality simulations are performed for selected years/scenarios (2016 baseline; 2035 and 2050 variants) due to computational demands. Ozone design values (DVs) and annual PM₂.₅ DVs are computed following U.S. EPA guidance by adjusting historical DVs with modeled relative responses.

• Electrification substantially increases electricity’s share of final energy and demand while CO₂ intensity of generation declines. Electricity’s share grows to 31% in 2035 and 34% in 2050 under High Electrification, and to 34% (2035) and 51% (2050) with a carbon price. Electricity demand is flat in Limited Electrification, but grows 23% by 2035 (39% by 2050) under High Electrification, increasing to 24% (2035) and 52% (2050) with a carbon price. Major new loads are transport (light- and heavy-duty) and industrial process heat. • Generation transitions away from coal in all scenarios; without national CO₂ policy, growth is met largely by natural gas and wind. With carbon pricing, coal is phased out earlier, much natural gas is retrofitted with CCS or co-combusted with hydrogen, nuclear remains, and solar/wind expand substantially. • Economy-wide CO₂ emissions decline across scenarios relative to 2005: −33% (Limited Electrification), −44% (High Electrification), and −78% (High Electrification + Carbon Price). Only the High Electrification + Carbon Price scenario aligns with the U.S. 2030 NDC (50–52% below 2005); other scenarios yield 19–28% reductions by 2030. • Criteria pollutant emissions fall most where fuel combustion is dominant. From 2016 to 2035: NOₓ declines 28% in Limited Electrification; 46% under High Electrification without carbon pricing; and 58% with carbon pricing. Under High Electrification + Carbon Price, EGU emissions drop dramatically relative to 2016: SO₂ −99% and NOₓ −82–87%. VOCs from on-road and off-road sources decline by >80% in High Electrification scenarios, partly offset by increases from oil and gas activities and other nonpoint sources. Primary PM changes little overall (+/−3–6%) and is dominated by fugitive dust; NH₃ increases ~20% (2035) and ~30% (2050) due to agricultural activity growth. • Ozone improvements are widespread and sizable. Deep NOₓ reductions (28–67%) yield large decreases in ozone DVs across CONUS, with greatest benefits in the Northeast, Southeast, and Ohio River Valley. 2035 Limited Electrification leads to attainment of the 70 ppb ozone NAAQS across the eastern states but not all western states due to higher background ozone. High Electrification (2035) reduces ozone by ~3–13 ppb; adding a carbon price yields additional reductions of ~6–10 ppb in the Midwest/eastern Texas and ~2–4 ppb elsewhere, driven by EGU NOₓ reductions (~75% nationally) and reduced oil and gas activity (~19% nationally). Urban areas see faster ozone declines than rural areas. Example monitors: Houston Aldine DV projected at 74 ppb (2035 Limited), 67 ppb (High Electrification), and ~63 ppb with carbon pricing; carbon pricing adds ~10 ppb benefit in northeast Texas vs. 2–4 ppb elsewhere. • PM₂.₅ benefits are modest and spatially variable. 2035 Limited Electrification reduces annual PM₂.₅ DVs by up to ~0.5 µg m⁻³ (e.g., −0.3 at Harvard Yards, OH) but increases occur at some monitors (e.g., Clinton, TX; Granite City, IL). High Electrification without carbon pricing reduces PM₂.₅ DV by ~1.0 µg m⁻³ at Harvard Yards; adding carbon pricing further lowers by ~0.5–1.0 µg m⁻³ (e.g., −0.7 at Harvard Yards). 2050 PM₂.₅ DVs are similar to 2035 with carbon pricing due to increases in primary PM₂.₅ (industry, dust) offsetting precursor reductions (NOₓ, SO₂). Secondary PM₂.₅ (sulfate, ammonium) and elemental carbon decrease; PM₂.₅ nitrate decreases in high electrification scenarios but can increase in winter in Limited Electrification in the Ohio River Valley and Texas. Modeling may overstate nitrate increases from SO₂ reductions, implying PM₂.₅ benefits may be understated. Organic aerosol decreases marginally; crustal material increases with primary PM emissions. • Method comparison shows simplified emissions-factor approaches understate CO₂ benefits of electrification. Compared to structural modeling (REGEN observed), constant marginal or average emissions methods capture only 52–91% of anticipated CO₂ reductions under current-policy electrification and 32–74% under carbon pricing, with larger biases when structural electric sector changes are substantial. Dynamic average rates approximate better than constant marginal rates but still miss decarbonization over time.

Electrification is an effective attainment strategy that delivers immediate, localized air quality improvements by 2035, complementing longer-term, global climate benefits. Deep NOₓ reductions—amplified by carbon pricing—drive large ozone DV decreases, enabling widespread attainment in the eastern U.S. However, PM₂.₅ presents a more complex challenge because not all components are combustion-related; interactions among nitrate and sulfate can attenuate benefits at lower control levels, and growing non-combustion sources (industrial primary PM, fugitive dust, agricultural NH₃) can offset gains. The linked US-REGEN–CAMx framework captures simultaneous structural shifts in end-use and power sectors and the nonlinear atmospheric chemistry essential for NAAQS-relevant metrics, revealing larger emissions and air quality benefits than implied by simplified marginal/average emissions approaches. Policy implications include the need to pair electrification and decarbonization with targeted controls on non-combustion sources to meet increasingly stringent standards and to consider regional heterogeneity in designing strategies.

The study demonstrates that end-use electrification, especially when paired with decarbonization policy, substantially reduces CO₂ and improves ozone and PM₂.₅ air quality across the U.S., with benefits that are immediate and geographically proximate. Structural modeling indicates that simplified marginal/average emissions factors underestimate electrification’s CO₂ benefits, particularly under strong CO₂ policies. While ozone improvements are large with deep NOₓ reductions, additional measures targeting non-combustion sources (e.g., industrial primary PM, fugitive dust, agricultural NH₃) are likely needed to achieve further PM₂.₅ reductions. Future research should evaluate alternative policy pathways to net-zero electricity and economy-wide CO₂, quantify community-level and environmental justice impacts, explicitly link air quality changes to health outcomes, and explore climate–meteorology feedbacks on emissions and air quality.

Air quality simulations use fixed 2016 meteorology, not accounting for climate-driven changes in biogenic or mobile source emissions under future warming. CAMx runs are limited to selected years and scenarios due to computational constraints. The study assumes a specific rising carbon price and does not explicitly model other proposed federal policies (e.g., clean electricity standards, mandates) or all potential pathways to net-zero. Some aerosol thermodynamics algorithms may overstate PM₂.₅ nitrate increases from SO₂ reduction, potentially understating PM benefits. Primary PM (e.g., fugitive dust) emissions are handled with simplifying assumptions (e.g., dust held constant except paved roads). The end-use model does not capture interactions among heterogeneous agents (e.g., peer effects). The analysis does not translate air quality changes into health impacts or address community-level environmental justice impacts. Broader macroeconomic feedbacks (e.g., CGE) are not included in this version.