Households in Washington, DC, help pay for the system that handles the rain falling on their roofs and driveways. DC Water bills every residential, multifamily and commercial customer a Clean Rivers charge based on how much hard, impervious surface their property has, and that charge funds the federally required program to cut sewer overflows into the city's rivers.

That's the context for an unusual bond DC Water sold in 2016. The utility wanted to try a different way of keeping rain out of the sewers, and it structured the deal so that investors would absorb part of the cost if the new approach fell short.

Why the rain is a sewer problem

Where Washington has a combined sewer system, stormwater and sewage run through the same pipes. In a heavy storm, the volume can exceed what the system can carry, and the mix overflows into the Anacostia and Potomac rivers and Rock Creek. Before the cleanup program began, according to DC Water, about three billion gallons a year overflowed this way.

The traditional fix is gray infrastructure — big tunnels and storage that hold the overflow until the treatment plant can take it. The alternative is green infrastructure: rain gardens, permeable pavement and planted areas that let water soak into the ground before it ever reaches a pipe. In 2016, DC Water says, it did not have much experience with green infrastructure, and it was considered a higher-risk bet than gray.

How the bond worked

In September 2016, DC Water sold a $25 million tax-exempt bond in a private placement to Goldman Sachs's Urban Investment Group and the Calvert Foundation. The money paid for Rock Creek Project A: 77 green infrastructure installations covering about 20 acres in the Rock Creek sewershed — rain gardens in planter strips and curb extensions, permeable pavement on streets and alleys, and two small parks designed to absorb water.

In most respects it was an ordinary municipal bond. The Environmental Protection Agency's April 2017 review of the deal describes a 30-year bond with an initial 3.43% coupon and a mandatory tender in year five — the point at which the performance would be settled. What made it unusual was a payment condition tied to how much stormwater runoff the project actually reduced:

  • Below 18.6%: investors would make a one-time $3.3 million "risk share" payment to DC Water.
  • Above 41.3%: DC Water would make a one-time $3.3 million "outcome" payment to investors.
  • Anywhere in between: neither side owed anything beyond the usual principal and interest.

The structure gave investors a financial reason to care whether rain gardens worked, and gave the utility some protection if they didn't. According to EPA, DC Water kept responsibility for designing, building and maintaining the project, and kept the risk of the interest rate when the bond reset. The investors weren't taking on construction risk. They were taking on a slice of performance risk, capped at $3.3 million on a $25 million bond — roughly 13% of the principal, by my arithmetic.

Measuring the result

The deal required performance to be measured, not estimated. EPA's review lays out a year of post-construction monitoring against a pre-construction baseline, a final report, and confirmation by an independent validator before any contingent payment could be made.

In 2021, DC Water reported that the project reduced stormwater runoff from the project area by nearly 20%, with the finding independently validated by the engineering firm WSP. That put the result inside the agreed range, so neither side owed the extra $3.3 million.

It's worth noticing where inside the range. Nearly 20% sits just above the 18.6% floor, and well short of the 41.3% that would have paid investors a bonus. The project worked, by the terms everyone agreed to, but closer to the low end of what the deal anticipated.

What the result doesn't tell us

The reduction was in runoff from one project area in one sewershed. It doesn't tell us flooding fell by 20% anywhere in Washington, and it doesn't establish that this kind of financing makes a project cheaper. DC Water still owed the usual principal and interest and still carried the cost of building and maintaining what it built.

It's also one deal. DC Water notes that Atlanta and Hampton, Virginia, have since issued environmental impact bonds of their own, but a handful of transactions isn't a track record.

And the most consequential outcome may not have been the runoff number at all. DC Water says the monitoring the bond required produced fifteen lessons about porous pavement, rain gardens and other installations — and that, based on those lessons, it plans a hybrid approach in the Rock Creek sewershed that blends gray and green infrastructure. By the utility's own account, the hybrid provides the same stormwater control as an all-gray approach at lower capital cost than either all-gray or all-green. In other words, the experiment informed a decision to build less green infrastructure than an all-green plan would have, not more.

So why would investors agree?

Why would investors accept the chance of giving up $3.3 million? They were paid interest on the bond, and had the chance to earn an extra $3.3 million if the project outperformed. The payments were symmetric. My guess is that, for investors focused on impact, it was also a way to put a price on whether an environmental intervention delivers.

For DC Water, the tradeoff bought something harder to get: a way to spread the cost of testing an untried method across a whole program. DC Water's own stated goals for the bond included shifting some performance risk to investors and improving transparency to ratepayers by publicly predicting, measuring and reporting the results.

My read is that the second goal is the underrated one. I wrote in What's Holding Up the Bond about how much of the municipal market rests on people paying a bill parcel by parcel. Many of those bills fund projects whose performance goes unpublished. Here, the households ultimately paying for the work got a number — measured, validated and public — and a contract under which some of the cost of disappointing results would have fallen on investors rather than entirely on them.

It didn't come to that. The project came in modestly above the line, and the cost stayed where it usually sits.

Sources

Figures current as of October 2026.