Effectiveness of Renewable Energy Policy for Air Pollution Reductions: Evidence from Wind Power in the US

Air quality improvement is frequently cited as one of the largest benefits of renewable energy (RE) policies. However, the air quality benefits of RE policy are seldom evaluated empirically after they have been implemented. Using hour-unit level data for 1666 electricity generating units (EGUs) in the US from 2011-2017, we examine how wind power development used to meet the Renewable Portfolio Standard (RPS) affected power sector emissions and air quality in the US. To do this, we combine quantitative data analysis with atmospheric chemistry modeling. We first calculate average marginal changes of unit-level electricity generation and emissions due to wind power, and find that wind power predominantly affects units using natural gas or sub-bituminous coal. We then estimate how wind power production used to meet the 2014 RPS target influences unit-level SO₂ and NO_x emissions, and project resulting changes in atmospheric PM_2.5 and O₃ concentrations using the GEOS-Chem model. We estimate reductions in CO₂, SO₂ and NO_x emissions of 41.4 million tons, 52.7 thousand tons and 26.4 thousand tons respectively, leading to a decrease in population-weighted PM_2.5 concentration in the eastern and central US in both January (-0.024 μg/m³) and July (-0.021μg/m³), but an increase in O₃ in January (0.035 ppbv). Monetized, the air quality co-benefits from avoided SO₂ and NO_x emissions are $867 million and the climate benefits from avoided CO₂ emissions (based on the social cost of carbon) are $791 million. We evaluate alternative policy scenarios that displace the same amount of electricity generation as the ex post scenario but with different criterion and illustrate how the estimating carbon and air quality benefits differ between scenarios. Alternative scenarios that maximize the avoided mortality from air pollution could deliver extra environmental benefits from avoided CO₂ emissions and air quality improvement ($1.1 to $3.3 billion), despite a small increase in cost ($0.16 to $0.26 billion).