What the water balance cannot see

Quoting your salary is not quoting your balance

If someone asks whether you are wealthy and you quote your monthly salary, you told the truth — just not to that question. Salary flows through you; the balance is what stays.

The Carpathian Basin is water-rich: the usual figure is roughly 11,000 m³ per person per year. That number is true, and it is a salary. Yet the basin grows drier every year — because a salary says nothing about what is left in the account.

This page is about how the measuring instrument itself decides what we notice. If we measure only throughput, then retention — a reservoir, a marsh, a recharged soil profile — leaves no trace on the balance at all.

What the national water balance actually measures

The Hungarian water balance is a good instrument: consistent, well researched, with decades of time series — one of the best documented systems in the basin. It tracks four items per year, in cubic kilometres:

All four items are throughput: how much water passes through in a year. For decades that was exactly the right question — what mattered was whether the water fitted in the channel. Today the question is how much stays here, and for how long. The balance cannot answer that, because it contains no item measuring the water that is actually here.

The same flows in and out — and the tub can still be empty

Picture a bathtub with water running in and the same amount running out. Throughput is constant. How much sits in the tub meanwhile is decided by one thing only: how long a drop stays before it leaves. Call it residence time.

Water retention — a reservoir, a marsh, a recharged soil profile — increases exactly this: it leaves throughput unchanged and raises what is held. That is why it does not show on the present balance, and why a second number is needed beside the four.

The difference of two large numbers, without an error bar

Domestically generated water — the water that originates here rather than arriving across the border — appears in the balance as the difference of two large numbers: 120 out minus 114 in leaves 6 km³.

Mathematically this is fine; in practice it is risky. It is like measuring two distances of about two kilometres each and deriving sixty metres from their difference: the measurement error can exceed the result itself.

One comparison says a great deal: between two publications by the same institution the large items differ by 16 km³ — almost three times the derived six. Neither states a reference period, so the reader cannot tell whether this is a decline or a methodological revision. Two things are missing from the number: which period it refers to, and what its error bar is.

Who decides about the water that arrives across the border?

Most of the water reaching us originates elsewhere. Over ten years the Danube lost roughly 10 and the Tisza 30 per cent of its discharge — the measurement is made here, the decisions behind it upstream.

When several parties share a resource and nobody enforces a limit, over-extraction is not a matter of intent but the natural outcome of the situation. The Murray–Darling reform worked in Australia because it fell under a single legislature: one cap, one court, one enforcer. The Carpathian Basin instead has bilateral agreements section by section, without a basin-level plan or authority.

Our own accounts are unsettled too: on some Tisza water bodies utilisation exceeds 100%, and the State Audit Office puts the number of unlicensed wells as high as a hundred thousand. In such a state a market instrument — water trading — would merely price uncovered demand. The order is fixed: measurement and a cap first, trading after.

The question "is it drying?" already has an answer

Satellite mass measurement has run continuously since 2002, NASA issued a warning for the region in March 2026, the OECD 2026 report shows a −3% annual trend in renewable freshwater, and the domestic model for the Nyírség finds more than 5 km³ of water lost and an average 102 cm drop in the water table between 2010 and 2022. This question is settled.

Once a question is answered, further measurement adds little — the bottleneck moves elsewhere. Here it moves to what exactly we are losing, whose loss it is, and what it is worth. Three items where this is missing today: the cost of the several-metre water table decline in the Homokhátság, to whom and over what horizon; the value of evaporation as landscape cooling; and the true price of network losses — the measured 24.5%.

This is no longer a measurement question but a choice of values: we have to state what counts as a loss. That is what turns measurement into decision.