Linking Policy to Statistical Uncertainty in Air Pollution Damages

Abstract

Abstract This study uses Monte Carlo analysis to characterize the uncertainty associated with per-ton damage estimates for 565 electric generating units (EGUs) in the contiguous United States (U.S.) This analysis focuses on damage estimates produced by an Integrated Assessment Model (IAM) for emissions of five local air pollutants: sulfur dioxide (SO2), nitrogen oxides (NOx), volatile organic compounds (VOCs), ammonia (NH3), and fine particulate matter (PM2.5). For each power plant and pollutant, the Monte Carlo procedure yields an empirical distribution for the damage per ton, or marginal damage. The paper links uncertainty in marginal damages to air pollution policy in two ways. First, the paper characterizes uncertainty in the magnitude of the marginal damages which is relevant to policymakers in determining the stringency of pollution controls. Second, the paper explores uncertainty in the relative damages across power plants. Relative damages are important if policymakers elect to design efficient regulations that vary in stringency according to where emissions are released. The empirical section of the paper finds that the marginal damage distributions are positively skewed and they are more variable for sources in urban areas than rural locations. The paper finds that uncertainty in three input parameters has the greatest impact on uncertainty in the magnitude of damages: the adult mortality dose-response parameter, the mortality valuation parameter, and air quality modeling. The analysis also finds that for each pollutant except for NOx only uncertainty in air quality modeling impacts efficient trading ratios calibrated to each firm's marginal damages.