Earth & climatePreprintExperiment3 min read

THE MOST EFFICIENT PRICE FOR CLEAN WATER: ZERO

Safe drinking water remains out of reach for most of Bangladesh. According to the national figures quoted in the paper, 61% of the population — around 100 million people — lack a safely managed source, 30 million use water contaminated with naturally occurring arsenic, and more than four in five drink water with bacterial contamination. In rural areas, the vast majority — 92% — rely on hand-pumped tubewells.

Piped water to every home would solve much of this. But someone has to decide what households pay to connect, and the decision seems to involve a painful trade-off. Charge more, and each paying household covers more of the bill; but fewer households connect, and the poorest are priced out. Economists call it the equity-efficiency trade-off.

A real pipeline, real prices

Md. Ahasan Habib, of the NGO Forum for Public Health in Dhaka, and Anna Tompsett, of Stockholm University, tested this in two communities of Bogra district, in north-western Bangladesh. They built an actual network: a deep borehole reaching an aquifer isolated from arsenic and bacteria, an electric pump filling a water tower, and pipes carrying water by gravity to private taps. It can serve about 200 households.

Asking people what they would pay rarely gives honest answers: some understate, hoping for a lower price, others overstate, hoping to attract the project. So the team used a method that rewards honesty. Each household stated the most it would pay; then it drew a sealed price from an envelope. If the drawn price was at or below its stated maximum, it could connect at that price — if its community won a public lottery for the pipeline. Overstating risked paying too much; understating risked losing the chance. A trial run with a bar of soap checked that households understood the rules.

Demand that falls off a cliff

The 330 participating households turned out to be very sensitive to price. The fitted demand curve gives 88% connecting at $1, but only 12% at $20; the median household would pay about $6.30. A provider seeking the most revenue would charge about $9.70 — and only a third of households would connect.

Now the costs. The network cost $4,795 in fixed costs — borehole, pump and tank — plus about $74 per connected household for pipes and taps. That fixed sum is spread over everyone who connects, so each extra household lowers the average cost.

Combining the two, the authors look for the price that minimises the public cost per connected household: what the network costs, minus the fees collected, divided by the number of homes connected. The answer is zero. At that price, everyone in the model connects and the cost is about $98 per household. Any positive price drives households away faster than it brings in money: at $9.70, the cost per connected household would rise well above that. Even a provider able to charge every household exactly its maximum willingness to pay would reach about $89 — just 9% less, before counting the cost of collecting fees.

The result is robust: willingness to pay would have to be at least 3.4 times higher than measured before charging anything became worthwhile.

What arsenic tests changed

The pipeline was built in the community that won the lottery, with the price eventually capped at about $8.40. 55% of households connected, close to the 53% predicted — and all 20 households that had drawn a free connection took it. But individual answers predicted individual choices poorly. Take-up depended instead on test results, shared after the survey: 40% where no arsenic was detected, 86% where it exceeded the national standard. Sources had been tested years before, yet the information had faded: 94% of households whose source had been tested believed it was safe, although only 8% of sources bore the green paint that marks a safe one, and their answers barely matched the later test results.

Two conditions

A simple model shows when free connections are the most cost-effective choice: large fixed costs, and demand that is very sensitive to even small prices — a pattern the authors note is widely documented among poor households. They acknowledge a single setting and a modest sample, and they leave open how such networks should be funded. Their warning to policymakers is precise: where economies of scale are strong — when large fixed costs make each extra connection lower the average cost — connection fees do not stretch a fixed budget further — they connect fewer households at a higher cost each.