Decomposition Analysis of the Role of Income and Policy on Water Pollution in the United States
Abstract
We investigate the effects of income and policy on water pollution in U.S. counties. Weestimate the effect of county changes in income and U.S. Clean Water Act and USDA Rural
Development grant awards to municipal water districts on water pollution outcomes.
Following Paudel & Crago (2020), we use watershed-level panel data on nitrogen,
phosphorus pollution, and dissolved oxygen readings for our water pollution measures.
These data are constructed for detailed US Geological Survey (USGS) hydrologic units
(eight-digit HU), approximately 2,200 geographic areas representing part or all of surface
drainage basins, a combination of drainage basins, or a distinct hydrologic feature. Water
quality data are from the Water Quality Portal maintained by the USGS, EPA, and
National Water Quality Monitoring Council. These data link to approximately 2,400 U.S.
counties for which income data are taken from the Bureau of Economic Analysis (BEA)
Gross Domestic Product (GDP) by County series and population from the Census Bureau
Local Area Unemployment Statistics (LAUS) series. Data on grants to municipal
wastewater treatment plants are derived from the EPA Grants Information and Control
System (GICS) (see Keiser & Shapiro 2018) and USDA Rural Development grant award
lists. To separate the effects, we employ a synthetic control method which combines elements
from matching and difference-in-differences techniques. It allows us to systematically select
comparison groups of counties and weight the control group to better match the treatment
group before the water abatement intervention. Thus, as usual with difference-in-difference
methods, we expect a tight pre-trend alignment between our control and treated counties,
followed by a divergence in water pollution and its determinants following implementations
of water pollution abatement policies.