Association for Public Policy Analysis & Management

In the wake of the Flint, Michigan tragedy, cities across the US are recommitting themselves to reducing risks of lead in drinking water. This paper focuses on the District of Columbia, which suffered its own high-profile water crisis in the early 2000s, and the efforts of DC Water, the District’s water utility, to remove lead pipes from the water distribution infrastructure. In particular, we model the probability of replacement of a lead service line (i.e., the pipe that connects a building to the water main under the street) using a unique dataset that matches household-level indicators from DC Water’s customer database and DC property tax files with census-tract-level indicators from the American Community Survey.

The equity dimensions of the issue naturally revolve around cost. The EPA’s original 1991 Lead and Copper Rule (LCR) required water utilities to pay the full cost of replacing lead service lines. Following legal challenges, a 2000 revision of the LCR allowed utilities to shift part of the cost to the property owner—namely the cost associated with replacing the portion of the service line on private property. At the same time, many cities proceeded with what are known as partial replacements, in which only the public side of the lead service pipe is replaced (from the property line to the water main) when the property owner declines to pay for the private side. DC ended its partial replacement program in 2008 as research began to indicate that partial replacements do not reliably reduce lead levels in water and may cause increased short-term exposure. A new DC law will begin offering free and discounted replacement of lead service lines on private property later this year, with the goal of reaching 100 percent full replacement.

Our paper focuses on patterns of replacement during the 2009-2017 period after partial replacements stopped and before subsidies kick in. Largely for reasons of affordability, we expect lower rates of full replacement for properties with lower assessed value located in census tracts with higher average markers of disadvantage; we will test that hypothesis using a regression model with neighborhood fixed effects. In addition, using data on the length of lead pipe on private property and replacement cost per foot, we will estimate price elasticity of demand for replacement and examine how it varies by income level and neighborhood demographics. From those estimates, the impacts of different levels of price subsidies on full replacement rates can be projected. Finally, using a difference-in-difference model, we will test the additional hypothesis that rates of full replacement will be higher for properties on streets where DC Water reached out directly to property owners to encourage private-side replacement as part of other capital improvement projects, such as repairs to, or replacement of, water mains.

The paper’s results are expected to be of interest to cities around the country that are exploring subsidy schemes, and for the EPA as it considers long-term revisions to the LCR.