Amman against Dubai: what a leakage percentage hides

Non-revenue water is the share of water put into a network that is never billed. It covers leaks, theft, meter error and unbilled legitimate use.¹ It is the headline efficiency number for any utility, and it is quoted as a percentage almost everywhere.²

On that measure Amman and Dubai sit at opposite ends of the world table. Jordan’s national utilities monitoring unit put the capital utility, Miyahuna, at 44.7 per cent in 2023.¹ DEWA reported 4.5 per cent for 2024 and 4.4 in a July 2026 update.³⁴ The obvious question is what Dubai knows that Amman does not. The answer is mostly that they are not measuring the same thing.

Amman does not run a pressurised network. It runs a rationing system. Miyahuna’s network is pressurised 22.6 per cent of the time, and 1.2 per cent of its subscribers receive continuous supply.¹ Households in Amman typically receive piped water two days a week and store it in rooftop tanks.⁵ Dubai reports smart water meters at 100 per cent of connections, more than a million of them by June 2024.⁶ It publishes no supply-hours indicator, and this case assumes continuous supply there.

Intermittency changes the arithmetic in both directions. A pipe that is empty most of the week leaks for fewer hours, which flatters the volume lost. Rationing can also suppress consumption, which shrinks the denominator and makes a given leak a larger percentage. In the worked example of a World Bank paper co-authored by Roland Liemberger, the first effect dominates. The same network, with no leak repaired, reads 20 per cent under 12-hour supply and 29 per cent once supplied continuously, because the volume lost doubles.²

Amman’s volumetric performance has improved, which the headline hides completely. The Utilities Performance Monitoring Unit of the water ministry and the Water Authority publishes losses per connection and per kilometre of main. Miyahuna’s loss per connection fell every year from 2019 to 2023, from 1.25 to 0.99 cubic metres a day. Loss per kilometre of main fell from 25.5 to 23.5. Over the same years the percentage went up, although the 2019 figure covers Amman alone.¹

The 2023 jump to 44.7 per cent has a stated cause, and it is not deterioration. The unit attributes it to the merger with Balqa. Excluding Balqa, whose own losses ran at 62.2 per cent, Miyahuna would have been at 38.9.¹ A utility that absorbs a worse network gets a worse number while doing nothing wrong.

The donor record in Jordan is poor, and the best account of it is an EU-funded policy review from May 2024 that tabulates the results without flinching. A KfW programme of EUR 18.8 million in Karak: no impact on non-revenue water, with the post-evaluation reporting an increase. A KfW programme of EUR 28.5 million in the northern governorates: little impact. A USAID programme of USD 42.5 million: gains not sustained, because operations staff breached the zonal boundaries the scheme depended on. The non-revenue water directorate built with EU support was dissolved in 2012.⁷ Cumulative donor commitments to the problem in Jordan reached about JOD 1.28 billion by 2024.⁷

That boundary failure is the instructive one. Zoning a network into measured districts only works if the operators respect the zones. The review found that operations units often violated zone boundaries in areas already rehabilitated.⁷ The incentive to open a valve during a shortage is stronger than the incentive to protect a measurement.

Dubai’s number deserves its own scrutiny, and the record does not fully support it. DEWA’s 2023 sustainability report and its statistics booklets give no water balance and no definition of what its network losses figure includes.⁸⁹ The booklets imply a gap between total water requirements and recorded consumption of 8.1 per cent in 2023 and 8.5 in 2024.¹⁰⁹ The published headlines for those years were 4.6 and 4.5.⁸³ Both pairs of numbers come from the same organisation.

One caveat about this section. No named independent analyst was found arguing that Dubai’s figure is not comparable with Amman’s. That argument is assembled here from the methodological literature applied to structural facts each utility publishes, and it should be read as reasoning rather than as a citation.

Stop comparing percentages across systems. A percentage is a ratio to the volume put into supply, and that volume is set by supply hours, pressure and demand, none of which are comparable between a rationed city and a continuously pressurised one. Losses per connection per day, read with supply time and pressure, survive the comparison.² On that basis the gap between the two cities is wider than the headline suggests, not narrower: roughly 77 litres per account per day in Dubai against about 990 per connection measured in Amman, and far more once normalised for the hours Amman’s pipes are actually full.⁹³¹

Read Dubai’s advantages structurally before treating them as management lessons. An almost entirely desalinated supply, complete smart metering and the revenue to maintain both are not choices a Jordanian utility can make.⁹⁶³ What does transfer is the measurement discipline, and Dubai has not published enough of it to be audited either.

Jordan is now trying a different instrument: a performance-based contract structured with the IFC, in which a private operator’s performance is measured by the volumetric reduction of non-revenue water.¹¹¹² The Ministry of Investment opened qualification in March 2026, and award is expected in 2026.¹³¹⁴ Given how many of the programmes behind that JOD 1.28 billion failed to hold their gains, moving the risk to whoever does the work is a reasonable next experiment. It will still fail if the zones are breached, which is a management problem no contract structure fixes on its own.