Date of Award


Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)


School of Public Health

First Advisor

Kai Chen



Background & Aims: Fossil fuel combustion emits a spectrum of ambient air pollutants, with the most harmful health effects seen from particulate matter 2.5 μm or less in diameter (PM2.5). Pricing carbon can reduce PM2.5 emissions and yield co-benefits for public health. The Regional Greenhouse Gas Initiative (RGGI) is a cooperative carbon pricing initiative among twelve states aiming to curb carbon dioxide emissions from the power sector. This preliminary analysis explores methodological approaches to quantify the health benefits associated with PM2.5 changes after RGGI was first implemented in 2009.

Methods: An interrupted time series analysis was performed to compare PM2.5 concentrations before and after the onset of RGGI compliance in nine RGGI states. Autoregressive Integrated Moving Average (ARIMA) models were selected using automated algorithms in R, applied for each state, and used to derive differences between the predicted and observed PM2.5 concentration values. The U.S. EPA Environmental Benefits Mapping and Analysis Program (BenMAP-CE) tool was used to quantify the PM2.5-related health impacts avoided due to the change in PM2.5 determined from the time series analysis.

Results: The ARIMA models selected by R demonstrated poor fit for the state-level time series data. Significant autocorrelation in the residuals remained, suggesting that some aspect or aspects of variability in the time series data were not adequately captured by the ARIMA models. Spatial correlation was suspected to explain some or much of the residual autocorrelation in the state-level time series data, but incorporating spatial elements was beyond the scope of this preliminary analysis. Despite poor model fit, state-specific intervention effect estimates plus an overall regional intervention effect estimate for the change in PM2.5 concentrations in the post-RGGI period were calculated for BenMAP-CE purposes. This change in PM2.5 was associated with avoidances in health impacts across all health endpoints studied.

Conclusion: For this preliminary analysis, ARIMA models demonstrated poor fit for the state-level time series PM2.5 concentration data. State-space ARIMA models or the incorporation of spatial lag variables into non-seasonal and seasonal ARIMA models may provide better fit in future iterations of this work. These models could inform future BenMAP-CE analyses across a range of health endpoints.

Open Access

This Article is Open Access

Included in

Public Health Commons