Monitoring Report: PNRR - Nobody's Laws

Romania is the European Union country where the most people die from treatable causes: diseases for which the diagnosis exists, the treatment exists and is funded, but the system fails to administer it in time. We invoke this statistic not to add yet another one to the pile of alarming statistics of these days. We invoke it because it most precisely describes the way in which the Romanian state fails: rarely through the absence of the correct diagnosis or the identification of the solution, almost always through the inability to administer it.

Full report (Romanian) - HERE.

And the PNRR (National Recovery and Resilience Plan) ends exactly like a preventable death — with the diagnosis made in 2021, the treatment fully funded, and the patient treated in the last shift, rushing through it, with half the doses. This is visible even in these very days: in the last week of July, one month before the final deadline, Parliament convenes in an extraordinary session to pass in five days the remaining PNRR reforms (the Urbanism Code, the integrity law, the decarbonisation of heating, the ANAF and Customs reward mechanism, the public servants' career reform), gasping after having turned a marathon into a sprint (the PNRR calendar has been available since 2021).

We put the thesis of this report on the table from the outset, so that it may be analysed — and because we hope to come out ahead at least in the treasury: Romania will very likely retain a large part of the PNRR funds and will fail regardless, because the money was never the real stake.

The PNRR was, in its letter, a reform contract with an attached investment budget. Romania executed it as an investment budget with some inconvenient reform conditions. And the failure that matters is not measured in euros lost, but in something else: the instrument built to increase state capacity was metabolised into a race against the clock, and the race against the clock into a blame game. In the end, the money remains and the same state remains. The Romanian state treats every deadline like an exam session: it studies the night before, forgets immediately after the exam, regardless of whether it passed or not. It has not yet understood the difference between studying to pass the exam and studying to be prepared for something.

The report tests this thesis from three directions.
First: what Romania talked about regarding the PNRR in the last three months. We analysed 13,756 public posts on Facebook, from the interval between the fall of the Bolojan Government and the eve of the extraordinary parliamentary session.
Second: what, in the same interval, the official implementation data show — 386 milestones and targets, 24,965 contracted projects.
Third: what five years of Factual checks on public statements about the PNRR show.

All three tell the same story, from different angles. At the end, we place the conclusions in a European context, because the Union has just decided to build half of its next budget on the PNRR model (the Multiannual Financial Framework 2028–2034).

I. What is Actually Being Voted on this Week

For those who do not follow the subject daily, the PNRR (National Recovery and Resilience Plan) is Romania's share of the European instrument through which the Union supported member states to recover from the COVID-19 crisis. Romania's plan was approved by the EU Council in October 2021, with objectives to be met by the end of 2026, on a detailed calendar by year and quarter. The money does not come freely. It comes in tranches, and each tranche is paid only if Romania ticks off a set of milestones (laws, reforms, institutions) and targets (kilometres, hospitals, installed capacities). Here lies the difference from classic European funds: if a milestone is not met on time, the payment request is blocked, and the money can be permanently lost. The final deadline of the entire mechanism, for all of Europe: August 2026. That is now.

One essential point, systematically forgotten in the public debate: this plan was not given to us. We wrote it ourselves. Each member state drafted its own plan, with its own priorities, within the limits of European targets (minimum 37% climate, minimum 20% digital). The milestones we have been talking about all summer are the promises Romania made to itself, in 2021, about what it needs to become a functional state. It negotiated them, signed them, and published them. Every time you hear a politician accusing Brussels of having forced us to carry out these reforms, that statement is false — and it is necessary to understand that the reform was written, negotiated, and assumed in Bucharest, by our own people.

The extraordinary session of July 27–31 was convened for the last milestone laws: the Urbanism Code, the integrity law, the decarbonisation of heating, the ANAF and Customs reward mechanism, the public servants' career reform (and additionally, outside the PNRR: the SAFE mechanism loan agreement). Payment requests 5 and 6 — well over 4–5 billion euros — hang on them. One milestone is, however, still missing from the list: the salary law, the most politically difficult reform of all. It did not make it onto the agenda because the (former) governing coalition did not reach an agreement on it. The question from which this analysis departs: if billions depend on these laws, what was Romania talking about during the three months in which they were supposed to be debated?

II. What We Measured and What We Found

We collected all public posts on Facebook, in Romanian, that touched on the PNRR subject between May 5 and July 24, 2026 (from the motion that brought down the Bolojan Government to the eve of the extraordinary session). The corpus contains 13,755 posts from 1,860 distinct pages. A Machine Learning (ML) model grouped them, with human verification, into fifteen general themes; for each post we retained not only the theme, but also how much reaction it gathered (likes, comments, shares) and whether the text is original or an identical republication of another message.

To compare themes with one another, we cannot look only at how many posts each has, because the number of posts and the attention they receive are two different things. We therefore use an attention index: how much a post from a given theme receives on average, relative to an ordinary post in the corpus. An index equal to 1 means that a theme receives attention proportional to how much is written about it. Above 1, the theme attracts more reaction than the number of posts would suggest. Below 1, the theme is written about a great deal, but read very little.

First finding: volume and reactions do not come from the same themes

The Economy is the most voluminous theme (44% of all posts), but its attention index is below one (0.89): an enormous amount is written about European funds, salaries, payment requests, but the public does not react proportionally. Domestic Politics has half the volume of the Economy (22.2%) and yet accumulates almost the same attention (index 1.70): every post about the crisis, the motion, and prime ministerial negotiations attracts 70% more attention than the average.

TOGETHER, THE TWO THEMES CONCENTRATE 66% OF POSTS AND 77% OF ALL ATTENTION — A CONVERSATION ABOUT MONEY AND POWER, NOT ABOUT PUBLIC SERVICES.

At the other end sit precisely the themes of concrete projects. Health (5.8% of posts), education (5.3%), governance (3.9%), and technology (0.6%) together account for 15.5% of volume, but only 4.8% of attention — each post among them receives between 15% and 35% of the attention of an ordinary post. A great deal is written about hospitals built on the final stretch, renovated schools, nurseries, and electric minibuses — and the public walks right past it. The small but heated themes behave exactly the opposite: censorship and transparency (index 2.35), social themes (2.97), and geopolitics (1.98) together account for under 1.2% of posts, yet every post among them gathers two to three times more reaction than the average. Conflict, accusation, and the frightening scenario circulate with visible intensity — but the concrete project is considerably absent from the debate.

One aspect relevant for reading this report is that part of the post volume does not reflect a real conversation, but the same message republished identically by different pages. In the governance and security themes, almost three quarters of posts (73% and 72%) are copies of a few dozen texts. In domestic politics, 42%. In other words, a significant portion of what appears to be debate is, in fact, a single message broadcast through a network of dozens of pages — a mechanism to which we return in chapter 3.

Second finding: the reforms being voted on now are absent from the conversation

The first finding is about broad themes. The second comes down to specifics: not "the economy" in general, but precisely the five reform laws that Parliament is voting on this week. We measured, across the entire corpus, two vocabularies. The first: the names of the laws on the session's agenda. The second: the register of blame and loss — keywords representative of the analysed topic: "sabotage," "betrayal," "deadlock," "motion," "political crisis," "we're losing the money."

The five reform laws appear, together, in 5.2% of posts and gather 7.1% of public attention. The blame register appears in 31.8% of posts and gathers 45.9% of attention. For every post that names a law, six name a culprit.

A few figures directly support the thesis. The verb "to lose," with its variants ("we're losing," "lost," "the loss of funds"), appears in 2,795 posts, and the word "billions" in 3,491. The five reform laws, taken together, appear in only 722 posts. The SAFE loan — a sum of money, with no reform attached to it in the debate — gathered more attention on its own than all five laws combined. The PNRR was thus visible to the public space as a figure and invisible as a set of reforms. The clearest example is the teaching career bonus: although it is on the Chamber's agenda in this very week, it appears in none of the 13,755 analysed posts — zero mentions in three months.

The mechanics of circulation say even more than volume. We divided the posts into ten equal groups, according to how much attention they gathered (we call these groups deciles: the first decile gathers the least-viewed 10% of posts, the last the most-viewed 10%). The reform vocabulary remains constant, between 4% and 7%, regardless of how widely the post circulates. The blame vocabulary, by contrast, rises from 30% in posts that almost nobody reads to 46% in the group of the most-viewed. Naming a law does not help a post circulate. Naming a culprit always does — and all the more so as the audience grows larger.

III. The ATM: A Defensive Strategy Against an Attack That Never Comes

Why does the political class talk about the PNRR almost exclusively in terms of sums? The explanation we encountered most often, in discussions in Bucharest and Brussels, is one of communication strategy: if the public discussion remains about money — billions coming in, billions that might be lost — then Eurosceptic politicians have no space to accuse an "imposition from Brussels." The Union becomes popular if presented, first and foremost, as a source of funding and, ultimately, as an ATM.

The data of this corpus show, however, that the strategy defends against an attack that never actually took place. The most frequent actors in the online discussions about the PNRR were the government and the institutions responsible for investments, while political actors from the Eurosceptic sphere were far less present than we would have expected. Sovereigntist-type criticism — conditionalities, dictates, blackmail from Brussels — is statistically almost absent precisely where one would have expected it to be loudest: the "EU and NATO" theme gathers 29 posts out of 13,756, while AUR pages published, together, only 27 posts about the PNRR in three months.

The dominant conflict was, instead, a different one. It pits the narrative of the reformist government, which "saves money and completes projects," against the narrative of political sabotage attributed to adversaries and, more marginally, against the critical narrative according to which the PNRR brings conditionalities, costs, waste, or the loss of sovereignty. The same actors — Ilie Bolojan, Sorin Grindeanu, Nicușor Dan, PSD, PNL, USR, the European Commission, the relevant ministries — change roles depending on the source: competent administrators in one set of posts, responsible for deadlock or corruption in another. In other words, the public did not argue over a narrative of European imposition of the PNRR, but over who loses the money — precisely on the terrain prepared by the ATM strategy.

Where did the anti-European narrative migrate, then? The Factual.ro archive (50 checks about the PNRR, from 2021 to today — see Annex 1 at the end) shows the trajectory. At the beginning, in 2021, we were mainly checking claims about allocations and consultancy costs — ordinary political debate. From 2023, however, the dominant category changes: "property rights would be conditioned through the PNRR," "the EU bans wood-burning stoves," "biometric identity cards will control us," "Romanians will lose their citizenship in 2031." The explanation is that the Eurosceptic narrative does not attack the PNRR directly as a political object — because that would mean denying the billions, and the billions are visible. Instead, it attacks its artefacts: the electronic identity card, the property register, the closed coal plant.

The conspiracy theory takes hold precisely in the void left where the public argument is absent and builds an imaginary that can attach itself to a very broad spectrum of fears present in society.

Here lies the fundamental error of the strategy described above. The absence of a public explanation of the role of European financing mechanisms — why they exist, what they reform, how they work — was left uncovered and fed the premise that "Europe means only money." And this void was occupied by the Eurosceptic sphere, which transformed it into its own narrative. If the Union is presented as a mere provider of funds, then any condition attached to those funds can be portrayed as a form of coercion. The only real defence against the "we're being dictated to from Brussels" narrative would have been the explanation of the reforms themselves: whose they are, what they fix, why we asked for them ourselves. That is precisely the conversation which, the figures above show, did not take place — because the idea persisted that a plan of such scope could be designed and implemented exclusively at the central level. We return to this point, at length, in chapter 5.

IV. The Plan We Wrote Ourselves and the Disease for Which We Prescribed Ourselves Nothing 

If the PNRR is Romania's promise about what it needs, then the way the money is distributed is a portrait of the state's priorities. The plan's six pillars are divided into 15 components, each with its own reforms and investments. The distribution of funds says a great deal about the angle from which the Romanian state looks at the need for reform and investment priority in Romania.

The green transition concentrates over 52% of the total budget (15.3 billion euros), and within it, the sustainable transport component, at 7.6 billion euros, is the best-funded in the entire plan. In second place is education, with 3.6 billion. At the opposite pole sits the "Good Governance" component — the reform of administration, of the public service, and of the way in which the state makes decisions — with under 1.5% of funds at the design stage. In execution, its share falls even further: 11 contracted projects, 80 million euros, 0.4% of the total value.

Here lies the central paradox: Romania is the EU state with the highest mortality from treatable causes — diseases for which diagnosis and treatment exist, but which the system fails to administer in time. It is, in other words, a problem of administrative capacity, and it is precisely to administrative capacity that the plan allocated the smallest portion of the entire budget.

This minimal allocation is not necessarily a miscalculation. Governance reforms are not inherently costly — a law on public servants' careers does not need billions. That is precisely where the real stake lies: the most important changes do not depend on budget, but on political will. This is worth viewing in a European context, not only a Romanian one.

The Draghi report on the Union's competitiveness (2024) identifies a series of "cheap wins" — low-cost but high-impact reforms, in the area of simplifying decision-making processes and governance. These are precisely the blockages that are not resolved with money, but with decision and political leadership. The relevance for the present report is that the weakness of governance is not a Romanian peculiarity, but a European problem — with Romania as an acute case. And the logic of "investment without reform of the way the state delivers" is, equally, a European model error, in which Romania functions as a case study.

The practical consequence of this logic is investment stripped of the reform that should have sustained it. Newly built nurseries in localities from which the young population has left — we observed this directly in election monitoring missions, where they often function as the best-equipped polling stations. Hospitals received, but without the staff that the salary reform — the milestone postponed even now — should have kept in the country. IT systems delivered to institutions that continue to work on paper. An investment without delivery reform is not a partial success, but an expenditure that misses its purpose, with all documents in order. Romania has traversed this pattern before, with the cohesion funds: rising absorption, far more modest institutional transformation. The PNRR had been conceived precisely as a correction of this pattern — conditioned payment was intended to ensure that investments could no longer advance without the transformation of institutions. In practice, Romania again separated the two components that the mechanism had bound together by contract.

V. The Parentless Project: Why Nobody Defends the PNRR

How does a country's plan that proposes (in theory) to reform itself end up with no supporters at all? The explanation begins with how it was designed. The public debate over the plan, in 2021, was minimal. Civil society, the business community, and local authorities sent proposals — many coming from practitioners, from people who know the terrain and feasible precisely for this reason. There is still no public explanation for why most of them were set aside.

We signalled this at the time and it is confirmed now: the consultation was another formality, not a negotiation or a real conversation. Then came the monitoring. The PNRR monitoring committee functioned more as an information channel (a sort of FYI, this will happen) than as a body with decision-making power: participation confirms presence, without influencing the plan's trajectory. And in the implementation stage, the involvement of society and the local level was almost entirely absent.

Why does this matter? Non-governmental organisations, together with local authorities and the business community, see reality on the ground: they know which investments work in practice, where the real blockages arise, and which reforms would need support to be accepted. A plan built with these actors gains corrections from direct experience and, more importantly, gains people and institutions willing to defend it when it becomes uncomfortable. A plan negotiated exclusively between the government and the Commission, implemented through emergency ordinances and inaugurated by mayors who had no power to intervene in its content, has, by contrast, nobody to support it.

The figures confirm this imbalance: 36% of the contracted value belongs to central beneficiaries, but they hold under 2% of the number of projects. Local authorities, on the other hand, execute thousands of small projects, decided and calibrated entirely at the centre. The local council submits the application, receives the funding, and inaugurates; the centre negotiates and sets the rules of the game. In the corpus, this balance of power is directly visible: the layer of 62% of local inaugurations (the nursery, the renovated school, the hospital wing "on the final stretch") is produced by the actors with the least decision-making power in the entire system — while the reforms, which are decided at the centre, are publicly defended by nobody.

The PNRR thus became a plan without supporters. Governments treated it as an inherited obligation, the opposition as an attack instrument, local councils as a funding counter, and civil society was kept almost entirely outside the process. When no actor assumes ownership, responsibility disperses — exactly the pattern from chapter 2: everyone accuses, nobody explains how and why it does (not) work. A plan that nobody owns ultimately produces reforms that nobody defends. There is here a deeper level, rarely discussed in the public debate: the acceptability of reforms.

A reform negotiated with an external evaluator and ticked off to unlock a payment does not go through an internal process of persuasion. Nobody builds a social majority around it, because the payment mechanism does not require this in the first instance. The consequence is that reforms made for the evaluator remain at the level of formal compliance, without being internalised into the practice of institutions. Romania has passed through this pattern before.

For almost a decade and a half, under the Cooperation and Verification Mechanism (CVM), it implemented reforms in justice, the rule of law, and anti-corruption externally monitored by the European Commission — and found at the end how fragile progress had remained where it had not been accompanied by a real substantive change and a real discussion in society with its citizens. The PNRR reproduced the same construct, this time with a payment calendar attached. The near-final result of the PNRR shows that strategic thinking about how funds and reforms should work together was absent not only during implementation, but already at the design stage, in 2021.

VI. The Reform Trated as a Tax on Investment: What the Implementation Data Show

The way Romania talked about the PNRR reflects how it implemented it. The official milestone data show a clear pattern. Reforms — predominantly legislative, i.e. exactly the type of measure that can be adopted rapidly through the procedural route, via the government's traditional emergency ordinances — are fulfilled at a rate of 73%. Investments — built schools, received hospitals, installed capacities, i.e. things that cannot be resolved through an ordinance — are fulfilled at a rate of only 34%. The difference says something about prioritisation: what could be ticked off quickly, through administrative means, was done first, and what required construction time, strategic thinking, and local-level debate to understand local needs was postponed.

The updated calendar confirms the same pattern. Of the 386 milestones and targets, 151 have a 2026 deadline, and 115 of these in the second quarter alone — which thus becomes the most congested quarter of the entire plan, placed right before the finish line. Of the 151, only 6 are currently assessed as fulfilled by the European Commission. The case of the Urbanism Code summarises this mechanism. The stage milestone — first reading — was ticked off in the fourth quarter of 2022. Entry into force, with a deadline in the second quarter of 2026, is only now being voted on, with Parliament recalled from recess, three and a half years later.

The July session is therefore not an exception to the way Romania has implemented the PNRR, but its characteristic form. As early as 2021–2022, approximately one third of the milestones requiring Parliamentary intervention were resolved through emergency ordinances adopted a few days before the deadline. Here a problem emerges that goes beyond the PNRR: the calendar of these obligations has been public since October 2021, but Romania tends to act through emergency ordinances adopted at the last moment, rather than through the ordinary parliamentary procedure. The result is that a deadline known for years ends up treated as an emergency — a symptom of an institutional culture in which emergency legislation has become a rule of operation, not an exception.

Where does the money lie, one month before the deadline

The contracted project database (PNRR Dashboard, updated at the end of June 2026) comprises 24,965 projects, worth 20.77 billion euros. The average technical progress, weighted by value: 54%. Financial progress: 33%. One in three projects (7,443) is at exactly 0% technical execution; together, these projects are worth 4.42 billion euros — 21% of everything contracted. 63% of value sits in projects below the 75% threshold. The private sector and research component has 90% of its value at zero — the financial instruments for companies have practically not started.

VII. The Parliament: The Final Decision at the Institution Without Stakes

In the final stage, the decision on reforms upon which billions depend arrives at Parliament. It is worth examining what kind of institution is being called upon to make it, in 2026. In recent years, Parliament's role in Romania has visibly eroded. From a space of debate and political confrontation based on representativeness — the place where community priorities and citizens' voices should meet — it has increasingly become an instrument at the parties' disposal, used predominantly for negotiation games among themselves. The adoption of legislation has slid away from the ordinary parliamentary procedure toward emergency ordinances and votes of confidence (the latter more frequently in the last government), subsequently ratified, and substantive debate has often been replaced by coalition calculations.

This institution, with its already weakened function, is now recalled from recess to adopt in five days reforms that were not built in five years.

It is worth explaining why Parliament had, in the structure of this mechanism, no real reason to behave differently. It had no significant role in the PNRR: it did not negotiate the plan, did not build its own monitoring committee for implementation, and did not request such a role. Governments fulfilled the legislative milestones predominantly through ordinances, which Parliament approved subsequently, sometimes a year later — in practice ratifying decisions already taken by the government. The result is a structure of incentives in which individual engagement does not matter. If the laws pass, the credit accrues to no particular MP. If they do not pass, the responsibility is collective and, precisely for this reason, diffuse. A body without ownership of the plan, without individual recognition for success, and without individual cost for failure cannot develop its own interest in the outcome. This is not an excuse, but a description of the mechanism. We formulate it this way because a mechanism of this kind — unlike simple indignation — can be corrected through rules.

VIII. The Part That Should Keep Us Up at Night: We Are About to Repeat Everything, at European Scale

Here the European Commission's proposal for the next multiannual budget enters — MFF 2028–2034. In essence, the current structure of 44 programmes is merged into 16. Fourteen currently separate funds — among them the cohesion funds that have financed motorways, hospitals, and water networks throughout Eastern Europe — merge into National and Regional Partnership Plans, negotiated by each capital with the Commission. And the financing logic changes: from cost reimbursement to payment conditioned on the achievement of milestones. In other words, exactly the PNRR model, extended to the scale of the entire budget.

The envelope of these national plans: 771 billion euros. 44% of the entire Union's budget for seven years.

It is worth re-reading everything that precedes with this perspective in mind. Relative to the size of its economy, Romania was one of the most extensive tests of the "milestone-conditioned financing, negotiated between capital and Commission" model. The results of this test are documented in the present report: reforms adopted formally, but without real public debate; investments completed at a rate of 54%; one fifth of funds at zero execution one month before the final deadline; administrative capacity financed at only 0.4%; local authorities and civil society maintained in a purely consultative role; a Parliament without its own role in fulfilling obligations upon which billions depended; and a public debate in which reforms remained practically invisible, responsibility was mutually attributed, and the anti-European narrative developed unhindered, around themes such as electronic identity cards or bans on wood heating. The model did not only produce incomplete execution. It also produced a certain type of politics — centralised, carried out under deadline pressure, lacking ownership of reforms, and accompanied by a constant dispersal of responsibility.

The conclusion drawn at European level from this type of experiment is the extension of the model to almost half of the budget. Here two observations emerge that deserve emphasis. The first: 27 separate national plans do not, in themselves, build a common European narrative, and the instrument that until now visibly linked the citizen to the Union risks being diluted into a negotiation between governments and the Commission. The second concerns the specific contribution of this data: the report provides the empirical part of the argument — concrete proof of what the model produces in a member state with reduced administrative capacity.

With the proposed extension, the Union is preparing to administer through this mechanism states with institutional capacities even more fragile than Romania's. The results of this first large-scale test do not appear to have been integrated into the debate about the future budget. Precisely because it was one of the most difficult cases, Romania has the most to say about how this model actually works — and has not, so far, formulated this position.

IX. What Could Each Actor Concretely Do

The results documented in this report do not describe an irreversible failure, but a set of dysfunctions that can be corrected. We formulate below a series of recommendations, by category of actor, each addressing a problem identified in the preceding chapters.

1. The Parliament: reclaiming the decision-making role

For the laws currently being debated in this session, one measure would bring immediate transparency: the publication, before the parliamentary debate, of a comparative table between the proposed form and the form negotiated with the Commission, article by article. A reform adopted in haste and modified without public debate raises greater risks than one that is delayed but openly discussed.

Beyond this session, we call for the establishment of a permanent monitoring committee for the implementation of European funds, with access to execution data and an obligation to publish periodic reports. Such a structure would give the legislature what was lacking in the PNRR: an ongoing, proprietary role in tracking a mechanism upon which billions depend. This is, in essence, the way in which Parliament can recover the power and decision-making representativeness it needs for the next financing instrument — and through which the chain of non-assumption of decisions, in which each actor passes responsibility to the next, can be broken.

2. The Government and MIPE: transparency of execution

A large part of the public confusion around the PNRR stems from the absence of a clear picture of the real status of projects. We recommend the publication of execution at milestone and project level, monthly and in open format, until the accounting closure of the plan. This would include the list of the 7,443 projects at 0% execution, together with the destination of each: continued from the national budget, transferred to other funds, or abandoned.

The 4.42 billion euros blocked in these projects require public clarification, project by project — either the confirmation that they are stopped, or a concrete completion plan. Such transparency would restore part of the information that is currently missing for citizens and would show, retrospectively, what worked and what did not — the minimum condition for knowing what can be improved in the future mechanism.

In parallel, it is worth addressing — and absolutely essential for the next plans, even belatedly — what weakened the plan from the design stage: the absence of real civil society participation. Building an effective consultation mechanism for the debate on the new MFF would be a first step in this direction.

3. The Romanian negotiators of the MFF (government, MEPs, the Presidency): capitalising on the PNRR experience

Romania is in a rare position: it has concrete data about what the milestone-conditioned financing model produces when applied to a state with reduced administrative capacity. This experience can be transformed into an argument in the negotiation of the new national plans regulation, so that it contains precisely the elements that were missing from the PNRR.

Particularly relevant in the European discussion are several points: mandatory territorial chapters, with their own budget; the real application of the partnership principle, in which local authorities, civil society, and the business community have decision-making power at the design stage, not merely a status as guests at an information event; dedicated tranches for administrative and technical capacity, paid at the beginning of the period, not at the end; and the full publication, in open format, of implementation data. For each of these demands, Romania holds the most solid empirical argument in the Union — its own trajectory.

Methodology and Data

1. How we collected the texts

We collected a corpus of 13,755 posts on Facebook in Romanian, disseminated in the period May 5, 2026 – July 24, 2026. Following dataset processing, we excluded 420 posts that did not fit any of the identified themes, some of them containing too little information to be interpreted. For thematic analysis we used a Machine Learning (ML) model for grouping similar texts, interpreted through a hybrid Human-In-The-Loop system based on NLP (Natural Language Processing) and RAG (Retrieval-Augmented Generation).

2. Thematic Analysis

For thematic analysis we used a Machine Learning (ML) model for grouping similar texts, interpreted through a hybrid Human-In-The-Loop system based on NLP (Natural Language Processing) and RAG (Retrieval-Augmented Generation).

The reach_proxy index, described below in this section, was used to measure audience. We summed this index of audience metrics (likes, comments, shares) for all unique messages belonging to a theme, obtaining the "Unique Proxy" score. In the case of multiple identical messages, we included the message with the highest values.

The "Total Proxy" score was obtained by adding all reach_proxy values for all texts belonging to a theme. The difference between "Total Proxy" and "Unique Proxy" reflects the audience obtained through verbatim-distributed messages. Thus, the Unique / Total ratio shows the percentage of audience obtained from original posts, out of the total posts.

3. How we calculated reach_proxy

In the absence of a direct measure of unique audience, a synthetic index of posts' amplification potential was constructed. Likes were assigned a weight of 2, comments a weight of 4, and shares a weight of 6. The weights reflect a progressive level of user engagement: liking involves a quick action and minimal effort, commenting involves formulating and expressing an opinion, and sharing directly contributes to extending the post's circulation to other user networks. The constant distance of two points between weights keeps the formula simple, transparent, and easy to apply to all posts.

The resulting value does not represent the actual number of persons who saw the post, but a comparative score of interaction intensity and propagation potential. As a sensitivity check, the indicator was also recalculated using equal weights for likes, comments, and shares. The hierarchy of the main themes was preserved, indicating that the comparative results are not determined by the choice of the above-mentioned weights.

4. Data concerning the PNRR and Factual.ro

Implementation data: the official list of milestones and targets with the evaluation status on payment requests (source: the European Commission's RRF dashboard); the contracted project database (source: the Ministry of Investments and European Projects' PNRR dashboard — PNRR Map — Vizualizare Fonduri PNRR România). Verification archive: factual.ro, 50 checks about the PNRR, 2021–2026 (Annex 1).

ANNEX 1 - Factual.ro Articles - HERE.