This is the repository for a project estimating the causal effect of the EPA's National Ambient Air Quality Standards (NAAQS) for PM2.5 level on transporation-related CO2 emissions in U.S. The project applies CausalArima (Menchetti et al., 2021) on EIA data on total transportation-sector CO2 emissions in the U.S. (1960-2023) as well as aggregated NASA data on on-road CO2 emissions at the county level in the contiguous U.S. (1980-2017). The preliminary results look promising.
data: This folder contains the "raw" data (downloaded online, internal to NSAPH, etc.) used in the analysis. If publicly available, the data sets are cited via links in footnotes.- Due to size constraints, the
CMS_DARTE_V2_1735(on-road CO2 emissions) anddataverse_files(PM2.5 concentrations) data sets are not included in this repository. - For
CMS_DARTE_V2_1735, see source or Google Drive. - For
dataverse_files, see source or Google Drive.
- Due to size constraints, the
files: This folder contains the files (.Rmd and .pdf) with the code and documentation.aggregation: This file aggregates data on PM2.5 concentration in the U.S. from the ZCTA level to the census block group level to join with the NASA data.final_county: This file performs causal inference at the county level. It first cleans the NASA data (outputtingco2_county.csv) and then runsCausalArimaon one specified county. Finally, withco2_county_causal_arima.csvfromrun_causal_arima.R, it plots the results of significant counties on a map of the U.S.final_national: This file performs causal inference at the national level. It first cleans the EIA data along with data on multiple potential covariates (outputtingnational.csv) and then runsCausalArimausing U.S. trade/GDP ratio, real GDP on a log scale, and urban population ratio as covariates.run_causal_arima: This file runsCausalArimaiteratively through every available county in the contiguous U.S. (withco2_county.csvfromfinal_county.Rmd) and outputsco2_county_causal_arima.csv.
plots: This folder contains the important visualizations generated in the files.results: This folder contains the important data sets generated, processed, and used in the files.