Network losses are not structural

There is a comfortable version of this topic in which Western Balkans network losses are high because the networks are old, the terrain is difficult and the region is poor. It is comfortable because it implies nothing can be done quickly.

The data does not support it.

Two neighbours, opposite directions

Across 2011–2024, on Eurostat's energy balance and one consistent definition:

• Serbia: 15.2% → 10.5%
• Albania: 14.0% → 17.3%
• Kosovo: 15.1% → 13.7%
• Montenegro: 15.4% → 13.9%
• North Macedonia: 14.7% → 14.1%
• Bosnia and Herzegovina: 9.3% → 9.4%

Serbia and Albania began the period within about one point of each other and ended nearly seven apart, moving in opposite directions. Bosnia and Herzegovina has been below 10% throughout.

If losses were a function of geography, climate or income, these six trajectories would not diverge like this. They diverge because losses respond to things governments and utilities decide.

What the indicator actually measures

Transmission and distribution losses combine two very different things, and conflating them is the main reason the number gets misread.

Technical loss is physics — resistance in lines and transformers, worse at low voltage, worse with long distribution runs, worse when the network is loaded beyond what it was designed for. Reducing it means investment: reconductoring, transformer replacement, voltage upgrades, reactive compensation.

Non-technical loss is everything else — unmetered connections, faulty or bypassed meters, unbilled consumption, billing failures. Reducing it means metering, enforcement and collection.

The two respond to entirely different interventions on entirely different timescales. A programme aimed at the wrong one produces expensive disappointment. Establishing the split is the first piece of work, and it cannot be done from an aggregate percentage.

Why the arithmetic matters here

Eurostat publishes losses in GWh, not as a share. The percentages above are derived — distribution losses divided by the electricity that entered the system (transformation output plus imports minus exports), computed identically for every country and year.

That is stated because the denominator is a choice. Dividing by gross generation, by domestic supply or by final consumption produces different percentages from identical underlying data, and comparisons that mix denominators are common and misleading. Both GWh inputs are kept alongside every point so the arithmetic can be checked.

The scale of a point

On Kosovo's 2024 volumes, one percentage point of loss is roughly 70 GWh a year.

For comparison: Kosovo's entire installed solar capacity of about 21 MW would generate somewhere in the region of 30 GWh annually. Closing a two-point gap recovers more energy than several years of solar build-out at the current rate — with no connection agreement, no land, and no permitting queue.

That is not an argument against building solar. It is an argument for costing loss reduction against generation on the same basis, which is rarely done.

What would change the answer

A current, disaggregated loss baseline — technical versus non-technical, by voltage level and by feeder. Kosovo's regulator publishes domestically; those figures are not yet ingested here, and this note relies on the Eurostat balance rather than on the national series. Anyone planning a loss programme should start from the regulator's own data, not from this.