Part I — Situation overview

At eight in the evening on 22 August the Prime Minister announced in a video message that alongside the two turbines operating until then at the Paks Nuclear Power Plant a third, turbine six of unit 3, would be switched on as well; around midnight the turbine reached three thousand revolutions and is generating electricity again. Károly Vigh, the plant’s operations director, described the course of the procedure on site: switching a turbine back takes forty-nine hours, of which the last three hours are the critical phase, when the rotation speed reaches the operating three thousand through checkpoints at six hundred and then nine hundred. The process itself is not exceptional: the specialists carry out between six and ten planned turbine restarts a year. According to the Prime Minister’s briefing the plant may again reach its full output of two thousand megawatts by Wednesday, and the next unit that can be started is turbine two of unit 1.

On the same day the Defence Minister, Romulusz Ruszin-Szendi, published a quantified progress report on the building of the Danube bottom sill — that is, the stone dam built into the riverbed to raise the water level: 231 metres have been completed on the left bank, and the work has reached the concrete blocks thrown in earlier. In total 3,895 cubic metres, that is 6,200 tonnes, of stone have been built in, and five civilian vehicles are being driven by soldiers between the quarry and the depot. The water level is rising, the sinking of the barges has been completed, and after the dismantling of the pumps the equipment will move over to the right bank. The government decided on the construction of the bottom sill on 12 August; since 2 August only the two turbines of unit 2 had been operating at full capacity at the plant, and from 10 August two turbines were generating again. According to HungaroMet’s communication, in the wider environs of the Alps — that is, in the catchment area of the Danube — precipitation exceeding fifty to sixty millimetres in places accumulated over three days, which is a reassuring development from the point of view of the river’s domestic water level; but no rain is still expected in Hungary, and in the second half of next week heat is coming again. According to the Prime Minister’s Sunday communication the water level of the Danube at Paks may fall further, and therefore Tuesday and Wednesday may be particularly important.

The story is not a Hungarian peculiarity. Both reactors of Romania’s only nuclear power plant, the Cernavodă facility, are down: unit one was shut down on 28 July, unit two on 13 August, because of the extremely low water level. On 21 August the Romanian disaster management committee decided to build a temporary dam on the Ostrov branch of the Old Danube, in order to direct a greater share of the little water available towards the Danube–Black Sea Canal and the plant’s water abstraction. In parallel, an average of fifty cubic metres per second more water was released from the reservoirs of the Olt river over five days, and alongside the dredging of the riverbed high-capacity pumps were also prepared. The army too was drawn into the work, with military divers and riverbed survey specialists. The two Romanian reactors together provide more than 1,400 megawatts.

MIAK’s reading of this is that the past two weeks are a combination of successful crisis management and a missing planning decision. The intervention worked: the water level is rising, the units can be switched back on, the logistics stood the test. At the same time the operability of a plant supplying a decisive part of the country’s electricity generation is being maintained by a river regulation element built in a few weeks on the basis of an individual government decision, and the flow of the river may fall again in the coming days. The question, therefore, is not whether the stone dam was a good idea, but how long low-water flow will continue to be called an extraordinary event, while in two consecutive summers it has limited the generation of both Danube-side nuclear power plants.

Part II — Foundations in the literature

The conceptual framework of the situation comes together in three authors. In his work The Economics of Climate Change — The Stern Review, Nicholas Stern (British economist, former chief economist of the World Bank, author of the 2007 government review on the economics of climate change) starts from the premise that the climate effects of the next two to three decades can no longer be avoided, so the only substantive answer is adaptation. Among its instruments he lists not emergency intervention but better information, better planning and climate-resilient infrastructure, with a quantifiable annual outlay. The central proposition of the volume This Time is Different by Carmen Reinhart and Kenneth Rogoff (Harvard economists, researchers who worked through eight centuries of financial crisis history) is “this time it’s different” thinking as a systemic error: decision-makers regularly see the recurring shock as an individual, particular case, and therefore do not price it in advance — the mechanism is not a financial peculiarity but a general planning error. In her work Governing the Commons, Elinor Ostrom (American political scientist, awarded the Nobel Memorial Prize in Economics in 2009 for her research on the management of commonly used resources) identifies eight design principles for the sustainable management of commonly used finite resources, among them the eighth, that of nested enterprises: for resources forming part of a larger system, appropriation, provision, monitoring and conflict resolution are organised in layers built into one another. In the case of the Danube this proposition directly describes what is missing: two states are building into the same river simultaneously, independently of each other. The detailed treatment of the literature — author by author, with quotations — can be found in section 6.4 Literature in detail.

Part III — MIAK’s concrete proposal

MIAK proposes three measurable measures. All three start from the premise that the acute phase of the crisis has closed, so now is the moment when the lessons can be written into rules — and not in the middle of the next low-water event.

3.1 A public hydrological N-1 criterion for the cooling water supply (by the first quarter of 2027)

MIAK asks that the domestic energy planning documents — the security of supply report and the network development plan — should contain a public, quantified operating boundary condition: at what Danube water flow and water level every unit of the Paks site is operable without separate riverbed intervention, and at which level the reduction of output, or shutdown, becomes mandatory. The N-1 criterion — that is, the yardstick that the system should work even if one of its elements fails — is an established method of network planning; the proposal is its hydrological equivalent, where the failing element is not a transmission line or a unit but a band of the river’s flow. The threshold has to be fitted to the low-water scenarios of the domestic integrated climate economy model (K9) and fixed within the framework of the energy market shock resilience programme point (K7). The practical benefit is that the publicity of the criterion also brings forward the moment of decision: if the flow forecast is heading towards the threshold, the system operator and the market know weeks earlier what additional capacity has to be booked — rather than at the moment when the shutdown is already an announced fact. Alongside the threshold value belongs the temperature limit on the return of the cooling water: the flow and the temperature together give the boundary condition, neither of them is sufficient on its own.

3.2 An itemised final report on the crisis management (within 60 days)

The second proposal concerns the accounting for the intervention now closing. MIAK asks that within sixty days from the completion of the intervention a public final report be published containing four items: the total cost of the bottom sill and of the sinking of the barges, broken down by source; the quantity of lost generation by unit and by period; the quantity and cost of the replacement electricity imports for the period of the shutdown; and finally a comparison of the flow model underlying the intervention with the measured rise in the water level. This is the joint application of the G1 data-driven budget programme point and the G25 energy price shock preparedness plan to a single, well-delimited event. The last item is the most important, because it is this that makes the technical decision verifiable: the difference between the rise in water level estimated when the bottom sill was designed and the actual rise shows how accurate the flow description of the riverbed is — and during the next low-water event that difference is already decision information. The report is not an after-the-fact criticism of the decision, but the input to the calibration of the threshold value under point 3.1: a threshold can only be responsibly established from measured data.

The third proposal seeks an answer to the question that, because of the nature of crisis management, nobody has so far asked: is the stone dam built into the bed of the Danube a temporary or a permanent structure. MIAK asks that the government take an express decision on this before the start of the next low-water season, and that the decision contain the procedural consequences that go with it. If the structure is permanent, then an accelerated procedure based on an emergency situation is not a sufficient legal basis: a full environmental impact assessment, a sediment and habitat study, and an assessment relating to navigability go with it — Danube freight traffic itself was faltering during the low water, so the effect of an element built into the riverbed on the shipping channel is a separate question. If it is temporary, then the criterion for removal has to be fixed: at what persistently measured flow and by what deadline it will be dismantled, who finances it, and what happens if it is needed again next summer. The absence of an exit criterion is the most expensive variant: the structure stays in the river for years without permit or review, because in no single year is the decision taken, always with reference to the following year.

The three proposals are held together by a single principle: the institutional response to a recurring event also has to be recurring, known in advance and verifiable. Reinhart and Rogoff’s observation about a recurring shock seen as an individual case (see 6.4.2) and Ostrom’s proposition on nested enterprises (see 6.4.3) lead to the same place from two different directions: a public, pre-established rule costs less than a solution improvised during a crisis, however skilful. And the common resource management framework under G28 adds that the low-water flow of the Danube is not a Hungarian internal affair: the Romanian parallel showed that two states are trying to direct more water from the same scarce stock to their own abstraction at the same time.

Part IV — Expected effects and risks

Dimension Expected effect Risk
Security of supply The public flow threshold brings forward the decision point; the system operator can book additional capacity weeks earlier The published threshold is also market information: the approach to the threshold may in itself raise futures prices, even before the actual shutdown
Budget The itemised final report shows the real cost of the shutdown, and with that the return on preventive investment becomes comparable The published cost will become the subject of political debate, which in the next crisis may create an incentive to withhold the report
Environmental protection Deciding the legal status of the structure forces the sediment and habitat study that was omitted in the accelerated procedure If the decision goes towards permanent maintenance, the study may degenerate into after-the-fact justification
Foreign policy Prior consultation on low-water riverbed interventions reduces the chance of parallel interventions cancelling each other’s effect The consultation obligation slows things down: in a crisis the notification deadline and rapid intervention may work against each other
Energy mix The cooling water risk appears as a planning parameter in the assessment of new capacity too Demonstrating water dependence may be simplified into a general argument against nuclear capacity, whereas the question is the choice of site and cooling technology

The main question for deliberation is the relationship between publicity and market effect. Publishing the hydrological threshold gives the market information about when a significant part of domestic generating capacity may drop out — and this information is built into prices, possibly even before the shutdown. In MIAK’s view this cost is nevertheless smaller than the cost of uncertainty: today what the market does not know is not how large the threshold is, but whether there is one at all. A hidden boundary condition does not eliminate the price effect, it only compresses it into the day of the announcement. The proposal tips over to the risk side if the threshold value is set but the model and the measured data behind it are not public: in that case the number becomes a basis of reference without being contestable. An ideology-free yardstick also belongs here: the final report now proposed makes the present government’s own crisis management measurable, including the possibility that the intervention was more effective than estimated.

Part V — Measurability and summary

5.1 What is worth following? (proposed KPIs)

The performance indicators below (KPIs, Key Performance Indicators) will show in 6, 12 and 24 months’ time whether crisis management has turned into a planning rule. These are proposed yardsticks, not government commitments.

  • Public threshold value: does a quantified Danube flow and water level threshold for the Paks site, broken down by unit, appear in the security of supply report by the first quarter of 2027.
  • Publication and content of the final report: is the itemised accounting produced within sixty days, and does it contain all four items — cost, lost generation, replacement imports, model deviation.
  • Model accuracy: how large is the difference between the rise in water level estimated when the bottom sill was designed and the actual rise; it is worth following whether this difference decreases at the next intervention.
  • Status of the structure: is the express decision taken before the next low-water season, and does it contain an exit criterion or a full impact assessment obligation.
  • Cross-border consultation: how many low-water riverbed interventions on the Hungarian and Romanian stretches of the Danube took place with prior bilateral notification, compared with all such interventions.

5.2 Summary

MIAK’s request to the decision-maker is three concrete steps: by the first quarter of 2027 a public hydrological threshold value for the Paks cooling water supply should be included in the official energy planning documents; within sixty days the itemised final report on the present crisis management should be published, comparing the model with the measured data; and before the next low-water season it should be decided whether the stone dam built into the bed of the Danube is a temporary or a permanent structure, and what goes with that in procedural law. And to the public, a request in outlook: the present restart is good news, but it is not the end of the story. The sentence “the intervention progressed faster than planned” is about crisis management; the sentence “at this water flow the unit is operable” is about planning — and of the two only the second protects us next summer.

Two MIAK foundational values are directly in play here. Data-drivenness, because the core of the proposal is making a single number public: the threshold value of the cooling water supply exists today within the expert system, but not for decision-making and market participants. And accountability, because an accelerated procedure based on an emergency situation by its nature switches off the usual control points — the itemised final report and the decision on the legal status of the structure are the two steps by which these can nevertheless take effect afterwards, without anyone calling into question the lawfulness of the rapid decision taken during the crisis.

Part VI — Justifications and further sources

6.1 The framing of the press, spectrum by spectrum

The liberal-left and public affairs band followed the event as a procedure, step by step. Telex was present with three pieces: one described the technical course of the restart of turbine six in the words of the operations director, the second carried the Defence Minister’s quantified progress report, and the third — from the transtelex editorial team — carried the Romanian parallel, that is, the interventions being made for the water supply of the Cernavodă plant. This is the only band that discussed the Hungarian and the Romanian event in the same news stream. 444.hu gave the most precise chronological frame: it recorded the shutdown of 2 August, the government’s decision of 12 August and also the sentence according to which the river may fall again in a week’s time. HVG highlighted the media form of the announcement: that the Prime Minister showed the switching back on in a live broadcast.

The economic band concentrated on measurable data. Portfolio put the figures of the progress report into the headline and the body text — the 231 metres, the 3,895 cubic metres, the 6,200 tonnes, the number of logistics vehicles — and on the same day published a separate piece on the work being done to keep the Romanian plant alive. The characteristic of the framing is that it measures progress as a quantitative indicator and does not turn the regulatory consequence into a question.

The conservative band highlighted the result of the intervention, not the process. Magyar Nemzet carried the completion of the sinking of the barges as its headline, and Mandiner that the bottom sill is being built, the water level is rising and generation may restart. On this day, therefore, all three bands carried the same set of facts, and the difference lies not in the selection of the facts but in the closing of the story: according to the conservative framing the matter has been resolved, according to the liberal-left one the reprieve lasts until the next fall in the water level. In MIAK’s reading this difference is precisely the question that proposal 3.1 puts into regulatory form.

6.2 Facts and data

Data Value Source
Completed section of the bottom sill on the left bank 231 metres Defence Minister’s progress report, 22 August 2026 (after Index, Portfolio, Telex)
Quantity of stone built in 3,895 cubic metres, 6,200 tonnes same
Duration of switching a turbine back on 49 hours, of which the last 3 hours are the critical phase Telex, 23 August 2026
Operating rotation speed of turbine six 3,000 revolutions per minute Telex, 23 August 2026
Full output of the Paks site 2,000 megawatts Telex, 444.hu, 22–23 August 2026
Expected attainment of full output Wednesday (26 August 2026) 444.hu, 22 August 2026
Number of planned turbine restarts in a year 6–10 Telex, 23 August 2026
The government decision on the bottom sill 12 August 2026 444.hu, 22 August 2026
Units in operation at the low point of the shutdown the two turbines of unit 2 (from 2 August 2026) 444.hu, 22 August 2026
Precipitation in the Danube catchment over three days exceeding 50–60 millimetres in places HungaroMet (Index, 22 August 2026)
Shutdown of the Cernavodă reactors unit 1: 28 July, unit 2: 13 August Agerpres, Nuclearelectrica (Telex, 22 August 2026)
Output of the Cernavodă plant more than 1,400 megawatts Telex, 22 August 2026
Additional flow released from the Olt reservoirs on average +50 cubic metres/second, over five days Romanian government communiqué (Telex, 22 August 2026)
Additional annual cost of making infrastructure climate-resilient in the OECD countries $15–150 billion, 0.05–0.5 per cent of gross domestic product (GDP) Nicholas Stern: The Economics of Climate Change

The last row of the table gives the post’s non-obvious conclusion. The Stern Review’s estimate — that the additional annual cost of making new infrastructure climate-resilient in developed countries is between 0.05 and 0.5 per cent of GDP — is important because it relates not to climate policy as a whole but expressly to adaptation built into planning. This order of magnitude gives the yardstick for the final report under proposal 3.2: if the combined cost of the present shutdown and the intervention — together with the price of lost generation and replacement imports — exceeds the sum from which a lasting solution for the cooling water supply could have been financed, then delayed adaptation has proved more expensive in numerical terms. This calculation, however, can only be carried out in possession of the itemised accounting; that is why MIAK asks for the final report, and not in order to pass judgment on the crisis management after the fact.

6.3 Policy dimensions

  • Environment and climate (programme points) — the cooling water risk as a planning parameter, and bringing low-water scenarios into the model: K7, K9, K2;
  • Economy (programme points) — the itemised accounting of the crisis management and the energy price shock preparedness plan: G1, G25, G28;
  • Foreign policy (background material) — the water use coordination of the Danube states, and the handling of the Romanian parallel;
  • Transport and infrastructure (background material) — the interaction of Danube navigability and the structure built into the riverbed, and the restoration of freight traffic after the low water.

6.4 Literature in detail

6.4.1 Nicholas Stern: The Economics of Climate Change — The Stern Review

Stern’s review regards adaptation not as a secondary but as an unavoidable task: there is no longer any instrument for preventing the climate change that will occur in the next two to three decades, but there is one for protecting societies and economies. For this the volume names four concrete instruments — better information, better planning, more climate-resilient crop varieties and more climate-resilient infrastructure — and it also estimates their cost.

“The additional costs of making new infrastructure and buildings resilient to climate change in OECD countries could be $15 — 150 billion each year (0.05 — 0.5% of GDP).”

The estimate is usable in the Paks matter because it relates not to the damage but to the price of preparation: it shows how much more expensive it is to design a facility for the new climatic boundary conditions from the outset. Stern also warns that for long-lifecycle capital goods — he expressly names power plants — today’s planning decision fixes the exposure for decades, and later correction is disproportionately more expensive. In the Hungarian situation this means that the cooling water risk is not an operational inconvenience of an existing facility but a planning parameter that also determines the capacity decisions of the coming decades — and that therefore has to be fixed in public, quantified form.

📖 Source: Nicholas Stern: The Economics of Climate Change — The Stern Review

6.4.2 Carmen Reinhart — Kenneth Rogoff: This Time is Different

The central proposition of Reinhart and Rogoff’s volume is “this time it’s different” thinking as a recurring decision-making error. According to the authors the period before crises is regularly characterised by the conviction that the old rules no longer apply, because the actors have learned from their mistakes, decisions are better founded, and circumstances are different.

“This brings us to our central theme — the ’this time is different syndrome.’ There is a view today that both countries and creditors have learned from their mistakes.”

The subject of the volume is financial crisis, but the mechanism is general: interpreting a recurring event as an individual case prevents the system from pricing it in advance. In the case of the Danube’s low water this pattern is clearly visible. The summer of 2026 is the second consecutive season in which the river’s water level has limited Hungarian and Romanian nuclear generation, and on both occasions the response was an extraordinary measure: an individual government decision, an accelerated procedure, military assistance. In the Reinhart–Rogoff framework this is not an error in the crisis management but a missing step before the crisis: a recurring event has to be written into a rule when the system is precisely not in crisis — that is, now.

📖 Source: Carmen M. Reinhart — Kenneth S. Rogoff: This Time is Different — Eight Centuries of Financial Folly

6.4.3 Elinor Ostrom: Governing the Commons

Ostrom identifies eight design principles in those common resource systems that have survived across generations. Of these the eighth relates expressly to those resources that are parts of a larger system — the Hungarian stretch of the Danube is precisely such a case.

“For CPRs that are parts of larger systems: 8. Nested enterprises. Appropriation, provision, monitoring, enforcement, conflict resolution, and governance activities are organized in multiple layers of nested enterprises.”

The essence of the principle is that for a resource forming part of a larger system, appropriation, provision, monitoring and conflict resolution are organised not on a single level but on levels built into one another: the local rule does not contradict the rule applying to the river as a whole, and vice versa. The present Hungarian–Romanian situation shows the absence of this. Two states are trying, in the same weeks and from the same scarce water stock, to direct more water to their own plant’s abstraction: Hungary with the stone dam built into the riverbed, Romania with a temporary dam on the Ostrov branch of the Old Danube and by drawing down the Olt reservoirs. The two interventions partly cancel and partly amplify each other’s effect — but since there is no common model and no prior notification obligation, neither party can calculate this in advance. In Ostrom’s framework this is not deliberate selfishness but the absence of the nested enterprise: the level at which the two interventions could be reconciled does not exist in decision-making form.

📖 Source: Elinor Ostrom: Governing the Commons — The Evolution of Institutions for Collective Action (Cambridge University Press, 1990)

6.5 International comparison

Over the past decade the French electricity system has several times limited the output of river-cooled nuclear power plants because of the low water level, or the high water temperature, of the Rhône and the Garonne. The essence of French practice is not the intervention but the rule: the river water temperature and flow thresholds are fixed in advance, the procedure and time limit of a temporary derogation are regulated, and the decision is foreseeable for system operation. This is precisely the structure that proposal 3.1 asks for: the difference is not that there is no water problem in France, but that the water problem is built into the planning.

On the side of Danube coordination, the existing frameworks of the International Commission for the Protection of the Danube River (ICPDR) provide the form in which the prior notification and joint modelling of low-water interventions can be interpreted. The present simultaneous Hungarian and Romanian riverbed interventions show that the existence of the framework is not enough: the bottleneck is not the institution but the obligation to notify an intervention in a way that is also calculable for the neighbouring state. This is the operational implementation of Ostrom’s eighth design principle on a specific river.

Environment and climate

  • K7 — Energy market shock resilience
  • K9 — Domestic integrated climate economy model
  • K2 — Energy transition plan

Economy

  • G1 — Data-driven budget
  • G25 — Energy price shock preparedness plan
  • G28 — Framework for common resource management

Proposed new programme point: Hydrological N-1 criterion for generating capacity with high cooling water demand — for the Environment and climate area.

6.7 List of sources

Press sources (MIAK press monitor, 23 August 2026 — topic 1):

Knowledge base references (literature):

  • 📖 Nicholas Stern: The Economics of Climate Change — The Stern Review
  • 📖 Carmen M. Reinhart — Kenneth S. Rogoff: This Time is Different — Eight Centuries of Financial Folly
  • 📖 Elinor Ostrom: Governing the Commons — The Evolution of Institutions for Collective Action

Note: the local file path of the books does not appear in the visible text of the blog — only the author and the title. The file path is an internal matter of the generation process, not the reader’s.

MIAK internal materials:

  • MIAK policy area: Environment and climate (programme points; programme point ID: K7, K9, K2)
  • MIAK policy area: Economy (programme points; programme point ID: G1, G25, G28)
  • MIAK policy area: Foreign policy (background material)
  • MIAK policy area: Transport and infrastructure (background material)
  • MIAK press monitor, 23 August 2026 — topic 1, score: 92/100

Supplementary public data sources:

  • HungaroMet — precipitation and catchment reports
  • General Directorate of Water Management — Danube water level time series
  • MAVIR — system load and import data
  • International Commission for the Protection of the Danube River (ICPDR) — Danube basin reports
  • Copernicus European Drought Observatory — drought index

Generation metadata