Accountability analysis of title IV phase 2 of the 1990 Clean Air Act Amendments

In this study, we sought to assess what portion, if any, of the reductions in ambient concentrations of particulate matter (PM*) < or = 2.5 microm in aerodynamic diameter (PM2.5) that occurred in the United States between the years 1999 and 2006 can be attributed to reductions in emissions of sulfur dioxide (SO2) and nitrogen oxides (NOx) resulting from implementation of Phase 2 of Title IV of the 1990 Clean Air Act Amendments. To this end, a detailed statistical model linking sources and monitors over time and space was used to estimate associations between the observed emissions reductions and improvements in air quality. Overall, it turned out to be quite feasible to use relatively transparent statistical methods to assess these outcomes of the Phase 2 program, which was designed to reduce long-range transport of emissions. Associations between changes in emissions from individual power plants and monitor-specific estimates of changes in concentrations of PM2.5, our indicator pollutant, were highly significant and were mostly of the expected relative magnitudes with respect to distances and directions from sources. Originally estimated on monthly data for a set of 193 monitors between 1999 and 2005, our preferred model performed equally well using data for the same 193 monitors for 2006 as well as for an additional 217 monitors not in the original set in 2006. Although substantial model uncertainty was observed, we were able to estimate that the Title IV Phase 2 emissions reduction program implemented between 1999 and 2005 reduced PM2.5 concentrations in the eastern United States by an average of 1.07 microg/m3 (standard deviation [SD] = 0.11 microg/m3) compared with a counterfactual case defined as there having been no change in emission rates per unit of energy input (1 million British thermal units [BTUs]). On a population-weighted basis, the comparable reduction in PM2.5 was 0.89 microg/m3. Compared with the air quality fate and transport models used by the U.S. Environmental Protection Agency (EPA) to estimate air quality improvements associated with the Clean Air Interstate Rule (CAIR) for 2010 and 2015, when baseline PM2.5 concentrations were expected to be about one-third lower, our statistical model yielded roughly similar results per ton of SO2 reduced, well within the estimated confidence intervals of the models. We have proposed a number of steps to advance air quality outcomes research using statistical methods. Specifically, we have emphasized the value of updating our analysis with post-2005 data to try to corroborate our findings. We have also recommended extending the work on air quality outcomes to include changes in health outcomes that might be associated with the implementation of Title IV Phase 2.