Part I — Situation overview

In the last days of July 2026 the water level of the Danube sank to a level never seen since measurements began. At Budapest the gauge stood at 30 centimetres on the morning of 28 July; at Paks minus 118 centimetres were measured on 29 July and minus 121 centimetres on 30 July. The Paks nuclear power plant obtains the cooling water for its reactors from the Danube, and returns the heated water there too — according to the environmental rule in force, 500 metres from the discharge point of the cooling water the temperature of the receiving river may not exceed 30 degrees Celsius, which in an extraordinary case, for reasons of security of supply and with the consent of the minister responsible for energy policy, may be exceeded up to 31.5 degrees. It was this statutory limit that wrote the week’s timetable: on Monday the output of Unit 1 was cut by 245 megawatts, on Tuesday evening that of Unit 3 by 237 megawatts, on Wednesday at 3 p.m. the complete shutdown of Unit 3 began, and from 7 a.m. on Thursday morning the output of Unit 2 was halved. According to MVM’s statement, the combined effect of the persistent drought and the heatwave has drastically reduced the Danube’s flow rate, and the derating of the units became necessary several times in June and July as well.

The governmental response ran along two threads. Minister of Economy and Energy István Kapitány emphasised that the shutdown is not a malfunction but a responsible measure taken on the basis of predetermined rules, and that the lost quantity is being replaced with domestic power plants, renewable generation and increased imports. In a video statement, Prime Minister Péter Magyar spoke of a 750-megawatt loss out of the 2,000 megawatts of installed capacity, and referred to the country having 3,600–3,800 megawatts of import capacity, so that “even a possible complete loss of Paks would not endanger the supply of the population.” He announced that the Energy Supply Operations Unit has begun work and is preparing a rotational crisis plan: this sets out which large public-sector and industrial consumers must temporarily reorganise or reduce their consumption if necessary, with hospitals, social institutions and essential public services excluded. According to Minister for the Living Environment László Gajdos, the water authority has placed four pumping pontoons and two floating cranes on standby on the Paks stretch of the Danube, where 55 specialists are on continuous standby; pumping is not needed for the time being. The domestic situation is not unique: in Romania both units of the Cernavodă nuclear power plant shut down for the same reason — Unit 1 on 28 July, Unit 2 from 10 a.m. on 30 July — amounting to roughly 1,400 megawatts of capacity, about a fifth of Romanian electricity generation.

In MIAK’s reading, the most important sentence is the one the government itself uttered: this is not a malfunction. That is precisely the problem. A malfunction is accidental, whereas a cooling-water limit is a rule — and the rule will operate in exactly the same way every drought summer. What dropped out here is not a faulty machine unit but the assumption that the output of the largest domestic baseload plant is independent of the weather. The Hungarian electricity system has therefore not lived through an unexpected event; it has revealed an unpriced climate exposure: the cooling-water limit bites precisely when the heat also drives up consumption, and regional import capacity — as the parallel Romanian shutdown shows — narrows in the very same hour when it would be most needed. The question is therefore not whether supply is secure today, but whether scenario-based capacity planning exists for the summer when water shortage, heat and a loss in a neighbouring country all arrive at once.

Part II — Literature foundation

Three mutually complementary frames offer themselves for interpreting the situation. In his work A hiány (Economics of Shortage, 1980), János Kornai (an economist of Hungarian origin, the author of the theory of the soft budget constraint; professor at Harvard between 1986 and 2002) described systematically what happens when decisions in an economy are steered not by prices but by quantity signals: in such cases actors ration, queue and resort to forced substitution, while reserve — in Kornai’s concept of slack — can be waste and safety at once, depending on the purpose for which it is held. The rotational crisis plan is precisely such a quantity instrument, so Kornai’s question is directly applicable: who decides, on the basis of what rule, and does the party affected know this in advance? The European Union’s Directive (EU) 2022/2555, the cybersecurity and resilience framework known as NIS2, provides the legal yardstick for this: it draws the energy sector — expressly including organisations engaged in nuclear electricity generation — into the highly critical category, and names business continuity, the management of backup systems and crisis management among the mandatory risk-management measures, that is, it regards preparedness not as good will but as a legal obligation. In his work Warming the World, William Nordhaus (the Yale economist who received the Nobel Memorial Prize in Economics in 2018 for his work in climate economics) formulates the methodological requirement needed to decide on the long-term technical response — for example a cooling-tower investment: a climate-policy decision must be based on the explicit comparison of costs and benefits, not on political momentum. The detailed treatment of the literature — author by author, with quotations — can be found in the 6.4 Literature in detail section.

Part III — MIAK’s concrete proposal

MIAK proposes three measurable measures which together replace case-by-case crisis management with regulated, predictable preparedness.

3.1 A mandatory water-shortage scenario in power-generation capacity planning (from the 2027 planning cycle)

MIAK proposes that the capacity planning of the electricity system contain a mandatory, annually reviewed water-shortage scenario in at least three bands: a calm year, persistent drought, and extreme drought together with heat. Each band should carry a numerical assumption for the available output at Paks, for domestic renewable and gas-fired reserve, and for regional import capacity — the latter expressly for the case when the loss occurs simultaneously in the neighbouring countries, as now in Romania. The assumptions and results of the scenarios should be public, because the value of a security-of-supply plan is given not by its existence but by its verifiability. To this end MIAK also proposes a daily-updated public dashboard: the Danube’s flow rate and water temperature, the available and the lost unit output, current imports and reserve capacity all in one place. This is the direct application of the K7 energy-market shock resilience and the G25 energy price shock preparedness plan: the threshold-based logic of the programme points works only if everyone can see the physical data belonging to the thresholds. Its legal basis is the business-continuity requirement of the NIS2 Directive (see 6.4.2).

3.2 Advance publication of the rotational curtailment regime and its compensation rule (within 60 days)

The rotational crisis plan is currently being prepared, and the government has given a verbal promise that hospitals, social institutions and essential public services will not figure among the consumers that may be curtailed. According to MIAK this promise deserves statutory form, and that before curtailment occurs. The published regime must contain: the pre-fixed order of the consumer categories concerned, the extent and maximum duration of curtailment, the objective technical thresholds for ordering it (at which water level and reserve level what is triggered), the deadline for notification, and what compensation a curtailed industrial consumer is entitled to. The reason for this is not formal: the economic effect of a quantity restriction depends on how far it can be planned for. A factory that knows at which threshold and how far in advance it will be notified can reorganise its shift pattern; a factory that does not know will accumulate reserves or postpone production — in Kornai’s description this is precisely the self-reinforcing shortage logic of a system built on quantity signals (see 6.4.1). A regime published in advance also narrows the space for discretionary decision, which is a transparency question: it should not emerge after the fact who was curtailed and who was not.

3.3 A cooling-water exposure audit and an open cost-benefit analysis for the technological response (by the end of 2027)

The third proposal prepares the long-term technical decision. MIAK proposes that a cooling-water exposure audit be prepared for every domestic power plant that relies on river cooling: how large an output reduction is to be expected at various values of flow rate and water temperature, and with what frequency, according to the climate models, do these values occur up to 2035 and up to 2050. Building on the audit’s results, the options for the technological response — a cooling-tower investment, the professional renegotiation of the thermal load limits, the partial replacement of capacity by other generation modes — must be compared in an open cost-benefit analysis, together with the investment cost, the value of the expected forgone generation and the environmental impact. This would be one of the first genuine applications of the K9 domestic integrated climate-economic model: the model is being built precisely so that such decisions should rest on calculation rather than on sentiment. The methodological yardstick is Nordhaus’s requirement (see 6.4.3): a regulatory or investment decision must not be severed from the benefit it is meant to achieve. The simultaneous halt of river freight transport shows that the boundary of the audit is not the plant fence: the replacement of the lost shipping capacity by rail is also an input to the priority matrix of KO4.

The three proposals are bound together by a single principle: climate exposure must be turned into a rule before it turns into a crisis. The cooling-water limit is not an unforeseeable event but the known legal consequence of a known physical relationship. Where the relationship is known, preparedness is not a virtue but an obligation — and the measure of preparedness is that the decision rules become public before the event, not after it.

Part IV — Expected effects and risks

Dimension Expected effect Risk
Economy A curtailment regime published in advance makes industrial production plannable, reduces the loss involved in an outage and the pressure to stockpile Public scenarios may in the short term worsen investors’ judgement of the system, before the investments improve the situation
Public administration Ordering tied to technical thresholds narrows the space for discretionary decision and makes crisis management verifiable after the fact An overly detailed, inflexible rule may slow things down in a genuine emergency; it needs a narrow, documented possibility of derogation
Environment and climate The exposure audit and the open cost-benefit analysis turn the thermal load question into a professional rather than a political decision The renegotiation of the thermal load limit on “security of supply” grounds may set a precedent for a lasting relaxation of the protection of wildlife
Transport Pre-contracted capacity for replacing the river outage by rail moderates the fragility of the supply chain Expanding rail capacity takes years; the bottleneck will still be there next drought summer

The main deliberative question is the tension between predictability and flexibility. A curtailment regime published in advance and tied to thresholds is valuable precisely because it takes the decision out of daily deliberation — but this same property makes it risky if reality falls between the thresholds. The proposal therefore works if the rule contains a narrow derogation path subject to justification and documentation, which must be publicly explained afterwards. The second sensitive point is the thermal load limit: the limit of 30 and, exceptionally, 31.5 degrees Celsius is not a bureaucratic obstacle but the protection of the river’s wildlife. If relaxing it becomes the most convenient answer, then we are buying security of supply at the expense of the ecological state of the Danube — and this trade-off is again a decision that must be taken in an open analysis, not in a night-time press release.

Part V — Measurability and summary

5.1 What is worth tracking? (proposed performance indicators)

According to MIAK the following performance indicators (KPIs, in English Key Performance Indicator) are worth tracking over 12–24 months:

  • Generation lost because of the cooling-water limit: the total annual thermally constrained output reduction in megawatt hours, broken down by unit and by cause — the aim being that the quantity should be comparable year on year and public, so that the trend of exposure also becomes visible.
  • The operation of the public dashboard: the daily-updated, machine-readable publication of Danube flow-rate, water-temperature and reserve-capacity data — the aim being routine launch before the 2027 drought season.
  • The existence of a curtailment regime published in advance: consumer order, threshold values and compensation rule fixed at statutory level — proposed deadline: 60 days.
  • Import cover alongside a simultaneous regional outage: how much import capacity is available in the scenario where there is a cooling-water limit in the neighbouring countries too — the aim being that this figure should appear in the capacity plan, not only the 3,600–3,800 megawatts projected for the normal situation.
  • The completion of the cooling-water exposure audit and the cost-benefit analysis: proposed deadline: the end of 2027, with public publication.

5.2 Summary

The key message is simple: MIAK asks that water shortage should be not an event but a planning assumption. Concretely, three things: a mandatory, public water-shortage scenario in capacity planning, a rotational curtailment regime published in advance together with a compensation rule, and an open cost-benefit analysis for the long-term technical response. None of them is expensive, none of them is ideological — all three are procedural questions, and all three are worth something only if they are ready before the next drought, not during it.

Two MIAK foundational values are in play on this topic. Data-drivenness, because the cooling-water limit is a physical relationship: measurable, modellable, predictable — and if a decision can rest on such data, then leaving it to political deliberation is irresponsible. And transparency, because a quantity restriction always creates winners and losers: whoever does not know in advance how the rule applies to them cannot plan, nor can they check afterwards whether the same yardstick was applied to them as to others. MIAK therefore asks the same yardstick of every government: crisis management is credible if its rules can be read before the crisis too.

Part VI — Justifications and further sources

6.1 The press framing by spectrum

The liberal-left and public-affairs band carried the story primarily in a technical-chronicle frame: Telex and HVG followed the precise sequence of the water-level data and the unit-by-unit output reductions (“Because of the further falling Danube the output of Unit 2 of the Paks nuclear power plant is being reduced”; “No stopping: the output of Unit 2 is also being halved at Paks”), and HVG was the one that also quoted the statutory thermal load limit — the rule of 30 and, exceptionally, 31.5 degrees Celsius. From the point of view of policy understanding, this detail is the most important item in the daily news, because it shows that the shutdown is not a technical failure but the enforcement of an environmental rule. HVG also lifted the story into a European context, with the parallel disruption of the Rhine and of river trade.

The economic band emphasised system-level exposure: Portfolio led not with the Hungarian units but with the complete Romanian nuclear shutdown (“Unprecedented energy emergency in Romania: the only nuclear power plant is shutting down!”), and reported in detail that the two units of the Cernavodă plant provide about a fifth of Romanian electricity generation, and that the preventive, voluntary complete shutdown is the first such event in the plant’s history. This framing yields the most precise Hungarian lesson: import capacity is not free capacity if the neighbour needs it in the very same hour.

The pro-government and conservative band chose the announcement-centred frame: Mandiner and Magyar Nemzet placed the prime-ministerial video statement and the announcement of the rotational crisis plan at the centre (“Péter Magyar has made an announcement about the Paks shutdown”; “Paks nuclear power plant: decisive days ahead, even a complete shutdown is not excluded”), and Magyar Nemzet lifted the scenario of a possible complete shutdown into its headline. 444.hu quoted the same statement from the side of the governmental claim (“According to the government not even a complete loss of Paks would endanger the population’s electricity supply”), that is, the framing here is not the physical process but the credibility of the governmental reassurance. In none of the bands did the question MIAK regards as the most important appear in any substantive way: is there a public, numerical scenario for the summer when the heat, the drought and a regional outage all arrive at once?

6.2 Facts and data

Datum Value Source
Danube water level at Paks −118 cm (29 July), −121 cm (30 July) statement of MVM Paks Nuclear Power Plant
Danube water level at Budapest 30 cm, the lowest since measurements began Telex, 28 July 2026
Unit 1 output reduction −245 MW (27 July) HVG, 30 July 2026
Unit 3 output reduction, then shutdown −237 MW (evening of 28 July), shutdown from 3 p.m. on 29 July MVM statement, HVG
Unit 2 output reduction −50% (from 7 a.m. on 30 July) MVM statement
Output lost with the shutdown of Unit 3 750 MW out of the 2,000 MW installed capacity prime-ministerial video statement, 29 July 2026
Hungarian import capacity 3,600–3,800 MW prime-ministerial video statement, 29 July 2026
Thermal load limit 500 m from the cooling-water discharge 30 °C, may be exceeded up to 31.5 °C on security-of-supply grounds environmental rule in force, quoted by HVG
Water-authority standby on the Paks Danube stretch 4 pumping pontoons, 2 floating cranes, 55 specialists on 24-hour standby statement of the Minister for the Living Environment, 29 July 2026
Shutdown of Cernavodă (Romania) Unit 1 on 28 July, Unit 2 from 10 a.m. on 30 July; approx. 1,400 MW, ~20% of Romanian generation Romanian ministry of energy, Nuclearelectrica; Portfolio
Previous complete drought shutdown at Cernavodă 2003 (when only Unit 1 was operating) Portfolio, 29 July 2026

The most important relationship in the table is not a single figure but the sequence: the output reduction took place in four steps within four days, alongside continuously deteriorating water levels, and every step was justified by an environmental limit value, not by a technical fault. The 750-megawatt loss alongside 3,600–3,800 megawatts of import capacity is indeed manageable in itself — but the simultaneous 1,400-megawatt Romanian loss also shows that this import capacity is under strain in several countries of the region on the very same days.

6.3 Policy dimensions

  • Environment and climate (programme points) — the cooling-water limit is the climate-dependent variant of energy-market shock resilience (programme point ID: K7); the cost-benefit question of the technological response belongs to the domestic integrated climate-economic model (programme point ID: K9); the long-term transformation of the energy mix is the subject of the energy transition plan (programme point ID: K2).
  • Economy (programme points) — the threshold logic of the energy price shock preparedness plan (programme point ID: G25) provides the frame for the curtailment regime and compensation; the industrial impact of the loss of river transport capacity touches the supply-chain chapter of strategic industrial policy (programme point ID: G9).
  • Transport and infrastructure (programme points) — the only scalable alternative to the lost river freight transport is rail, so the transfer demand of the drought months must be added to the inputs of the investment priority matrix (programme point ID: KO4).

6.4 Literature in detail

6.4.1 Kornai: Economics of Shortage

Kornai’s central observation in A hiány is that decisions in an economy can be steered by two kinds of signal: the price signal and the quantity signal — and if the second gains the upper hand, the entire logic of adjustment changes. In Kornai’s formulation, “on both the output and the input side, the ‘quantity’ signals relating to shortage and to future demand have a far greater effect than economic and profitability calculations performed with the use of prices.” From this follow the familiar reflexes of the shortage economy: actors accumulate reserves, buy in advance and resort to forced substitution, because they regard access, not price, as the scarce thing. Kornai’s conceptual distinction concerning reserve also belongs here: he uses the English word slack because there is no neutral Hungarian equivalent — “surplus” suggests waste, while “reserve” sounds commendatory, even though the same unused resource can be both “good” and “bad” depending on the decision-maker’s aims. In the present Hungarian situation this frame says two things. On the one hand, the rotational crisis plan is by its nature a quantity instrument: it reduces consumption not through price but by designation. On the other hand, import capacity and reserve are slack — their value depends on whether they are part of a scenario thought through in advance, or whether we merely invoke that “in principle it exists.” The lesson is not that quantity restriction can be avoided, but that if we apply it, we should do so by rule, in advance and predictably — otherwise the consumers affected will take on the very defensive reflexes Kornai described in the shortage economy.

📖 Source: János Kornai: A hiány (Economics of Shortage, 1980)

6.4.2 European Union: NIS2 Directive

The significance here of Directive (EU) 2022/2555, known as NIS2, is not its cybersecurity subject matter but its legal structure concerning preparedness. The directive draws the energy sector — in its annex expressly the players of the electricity industry, and in its preamble separately the organisations engaged in electricity generation from nuclear power plants — into the highly critical category, with the reservation that in respect of activities relating to national security, member-state responsibility remains. And in the enumeration of risk-management measures, “business continuity, such as backup management and disaster recovery, and crisis management” figures as a separate point. The directive’s reasoning also records that because of the interdependencies of energy, transport, drinking-water and waste-water services and healthcare, a disruption originally confined to a single entity can also have cascading, ripple effects. In the Hungarian case this structure is directly applicable: according to the logic of EU law, continuity and crisis-management planning is not a matter of occasional governmental decision but a lasting obligation of the operator and the regulator — that is, the operations unit now set up and the rotational plan under preparation are not an extraordinary measure but the belated fulfilment of an already existing obligation. The proposal therefore does not ask for a new burden but for the public, verifiable filling-out of the existing framework.

📖 Source: European Parliament and Council: Directive (EU) 2022/2555 (NIS2)

6.4.3 Nordhaus: Warming the World

The methodological core of Nordhaus’s book is the cost-benefit comparison of climate-policy decisions: the model family (DICE and RICE) was built precisely so that “the costs of slowing climate change” could be set against “the benefits of slower climate change.” Nordhaus sharply criticises regulatory decisions that sever this connection — of the Kyoto Protocol he writes that the emission limits were introduced “without any attempt” to link them to the benefits of lower emissions. At the same time Nordhaus also points out that the early impact estimates — especially those relating to agriculture and energy — were excessively pessimistic because they did not reckon with adaptation; later studies that also model adaptation paint a less dark picture. Translated to the Hungarian cooling-water question, this tightens things in two directions. On the one hand, adaptation is a genuine possibility: a cooling tower, the review of the thermal load rule or the partial replacement of capacity is not a theoretical option but a calculable investment. On the other hand, precisely for that reason it must actually be calculated: the decision can be born from the explicit comparison of the investment cost, the value of the generation expected to be forgone and the environmental impact, not in the communication pressure of a drought week. Programme point K9 adapts this model family to Hungarian data precisely so that such decisions should have a domestic calculation basis.

📖 Source: William Nordhaus: Warming the World — Economic Models of Global Warming

6.5 International comparison

The climate exposure of a power-plant fleet built on river cooling is a documented and recurring phenomenon in Europe. In France, heat and low river levels led in several summers to the derating of nuclear capacity or to the issuing of temporary thermal load concessions; on the German stretch of the Rhine, the low water levels of 2018 and 2022 obstructed river freight transport and with it the supply of industrial raw materials — in the present Hungarian situation the same is repeating on the Danube and the Drava. The present preventive complete shutdown of the Romanian Cernavodă plant is a new precedent in so far as it was not a technical fault but the operator’s voluntary decision built on the hydrological forecast: the plant placed the preservation of nuclear safety margins ahead of generation. This decision logic is a transferable model — not the shutdown as an aim, but that the forecast concerning water shortage should be a formal input to the operating decision. The lesson, in summary, is what Nordhaus’s cost-benefit frame also suggests: the technological questions of adaptation are decided well by those countries that pose them not in crisis weeks but in a planning cycle — and that also count in regional simultaneity, because a Central European drought does not stop at the border.

Environment and climate

  • K7 — Energy-market shock resilience
  • K9 — Domestic integrated climate-economic model (DICE-HUNGARY)
  • K2 — Energy transition plan
  • K1 — Measurable climate targets

Economy

  • G25 — Energy price shock preparedness plan
  • G9 — Strategic industrial policy

Transport and infrastructure

  • KO4 — Rail development with data-based priority

Proposed new programme point: Water-shortage scenario in power-generation capacity planning — a mandatory, annually reviewed set of scenarios working with public assumptions, together with an associated curtailment regime published in advance — for the Environment and climate and Economy areas.

6.7 List of sources

Press sources (MIAK press monitor, 30 July 2026 — topic 1):

Knowledge-base references (literature):

  • 📖 János Kornai: A hiány (Economics of Shortage, 1980)
  • 📖 European Parliament and Council: Directive (EU) 2022/2555 (NIS2)
  • 📖 William Nordhaus: Warming the World — Economic Models of Global Warming

Note: the local file path of the book does NOT appear in the visible text of the blog — only the author and the title.

MIAK internal materials:

  • MIAK policy area: Environment and climate (programme points; programme point ID: K7)
  • MIAK policy area: Economy (programme points; programme point ID: G25)
  • MIAK policy area: Transport and infrastructure (programme points; programme point ID: KO4)
  • MIAK press monitor, 30 July 2026 — topic 1, score: 90/100

Additional public data sources:

  • Statements of MVM Paksi Atomerőmű Zrt. — unit-by-unit output data
  • Statements of Nuclearelectrica and the Romanian ministry of energy — data on the Cernavodă shutdown
  • HungaroMet water-level and water-temperature data, hydrological data publication of the National Directorate of Water Management
  • Copernicus European Drought Observatory — drought monitoring

Generation metadata