Machine learning predicts which rivers, streams, and wetlands the Clean Water Act regulates.

We assess which waters the Clean Water Act protects and how Supreme Court and White House rules change this regulation. We train a deep learning model using aerial imagery and geophysical data to predict 150,000 jurisdictional determinations from the Army Corps of Engineers, each deciding regulation for one water resource. Under a 2006 Supreme Court ruling, the Clean Water Act protects two-thirds of US streams and more than half of wetlands; under a 2020 White House rule, it protects less than half of streams and a fourth of wetlands, implying deregulation of 690,000 stream miles, 35 million wetland acres, and 30% of waters around drinking-water sources. Our framework can support permitting, policy design, and use of machine learning in regulatory implementation problems.

The low but uncertain measured benefits of US water quality policy.

US investment to decrease pollution in rivers, lakes, and other surface waters has exceeded $1.9 trillion since 1960, and has also exceeded the cost of most other US environmental initiatives. These investments come both from the 1972 Clean Water Act and the largely voluntary efforts to control pollution from agriculture and urban runoff. This paper reviews the methods and conclusions of about 20 recent evaluations of these policies. Surprisingly, most analyses estimate that these policies' benefits are much smaller than their costs; the benefit-cost ratio from the median study is 0.37. However, existing evidence is limited and undercounts many types of benefits. We conclude that it is unclear whether many of these regulations truly fail a benefit-cost test or whether existing evidence understates their net benefits; we also describe specific questions that when answered would help eliminate this uncertainty.