Questionable Lead Reporting for Drinking Water Virtually Vanished after Flint Water Crisis, Study Reveals

Public water systems in the U.S. were far less likely to report suspiciously rounded lead levels after the Flint, Michigan water crisis drew national outrage and federal scrutiny, according to new research led by the University of Massachusetts Amherst.

The study, published as the first article in the latest issue of American Economic Review: Insights, introduces new statistical methods to distinguish between natural rounding and potential "threshold manipulation" in reported figures.

"Existing methods can mistake rounding for manipulation," explains Tihitina Andarge, assistant professor of resource economics at UMass Amherst. "Our approach allows us to separate the two."

Andarge, David A. Keiser, professor of resource economics at UMass Amherst, Dalia Ghanem of the University of California, Davis, and Gabriel E. Lade of The Ohio State University analyzed how water systems reported lead concentrations from 2011 to 2020 under the Lead and Copper Rule, a key provision of the Safe Drinking Water Act. The rule requires systems to determine whether the 90th percentile of the lead concentrations in their water samples exceeds federal thresholds that can trigger additional monitoring, remediation and public notification.

The Environmental Protection Agency (EPA) relies on self-reported figures from about 50,000 water systems nationwide. Systems with reported lead levels above 0.005 milligrams per liter must continue frequent testing, while those above 0.015 must take costly corrective steps and notify the public. This creates an incentive for systems to report values just under the cutoffs.

The study found that before the Flint crisis prompted a state of emergency in 2016, about 3% of medium-sized systems and about 0.5% of small systems reported lead concentrations rounded exactly to the federal threshold-a pattern the researchers say is statistically unlikely to occur by chance. After Flint, those suspicious clusters all but vanished, and reported data aligned more closely with expected distributions.

Andarge notes that among small water systems, this pattern was concentrated in Alabama, while among medium-sized systems, it appeared throughout the country, though at a smaller scale.