On 24 August 2026, Manfred Weber and Valérie Hayer wrote to Ursula von der Leyen asking the Commission to make the disbursement of Romania’s final tranche of €770 million under the Recovery and Resilience Facility conditional on the fate of a national statute. Their legal argument occupies a single sentence: Dominic Fritz, the mayor of Timișoara, is a German national elected in Romania in the exercise of the rights conferred by Article 22(1) TFEU, his mandate is “therefore held by direct operation of Union law”, and its termination is therefore a matter the Commission may enforce. The letter arrived a week after the Romanian Constitutional Court had ruled on that statute in Decision 883 of 17 August 2026, published in the Official Gazette on 21 August, and at the point of maximum leverage: the milestone deadline of 31 August.

The claim I want to make is narrow and has two parts. First, the letter is right that something is wrong with Decision 883 and wrong about what. There is a serious constitutional objection to the decision, the three dissenting judges make it, and it is not the objection the letter makes; the Article 22 construction has no basis in the Treaty or in the case law, and it would, if accepted, do more damage to the Union legal order than the decision it seeks to correct. Second, and this is the part that outlasts the case, the letter argues in a register that cannot be argued with. It treats a contested and, on one point, badly reasoned judgment as the output of a captured institution. Capture is not a rhetorical intensifier. It is a proposition about how an institution actually works, and it can be tested. On the evidence available, Decision 883 fails that test. What remains is a decision one disagrees with, which is a different object, and which calls for a different instrument.

The distinction is not a matter of manners. It determines which instruments the Union reaches for now, and which ones it will still be able to reach for later.

What the Court Actually Held

The statute under review amends Romania’s integrity legislation and was itself an RRF milestone. Two of its features were challenged, by twenty-seven senators of USR and PNL and, separately, by President Nicușor Dan. The two must be kept apart, because the Court treated them differently and because the dissenting judges object to each on an entirely different ground.

The first is the sanctioning regime. Article II point 17, which rewrites Article 25 of Law no. 176/2010, together with the transitional provision in Article X, attaches a new consequence – immediate termination of the office, dignity or elected mandate – to a conflict of interest that had been definitively established before the new law entered into force. Under the previous regime such a finding carried a disciplinary sanction and a three-year bar on holding public office in the future; it did not end a mandate in course. On this point the objection was rejected, by majority.

The second is transparency. Article II points 2 and 5 extended the obligation to file, and to publish online, declarations of assets and interests to “the person who has relations similar to those between spouses”. Here the Court struck the law down: unanimously as to that phrase, unanimously as to one of the published categories of data, and by majority as to the remaining forty-two.

The separate opinion is signed by Laura-Iuliana Scântei, Dacian-Cosmin Dragoș and the president of the Court, Elena-Simina Tănăsescu. It dissents from the last two points of the operative part, on two unrelated grounds. On the sanctioning regime it argues that the majority is wrong about non-retroactivity. On transparency it argues that the Court should not have ruled at all, because it extended its review, of its own motion, to provisions neither the President nor the senators had challenged, which Article 146(a) of the Constitution does not permit. The letter presents the three as a bloc of dissenters standing against a compliant majority. They are not a bloc: one of their two objections would have restricted the Court’s own reach rather than extended it.

The Objection Worth Making

Article 15(2) of the Romanian Constitution prohibits retroactivity in absolute terms, with the single exception of the more favourable criminal or administrative-offence law. Romanian constitutional doctrine has never developed a graduated theory of retroactive effect comparable to the German distinction between echte and unechte Rückwirkung. The prohibition either applies or it does not, which means the entire weight of the analysis falls on how the facts are characterised.

The Court characterised them by separating the office from the person who holds it. On that reading the law does not sanction an individual for past conduct; it defines, prospectively, the conditions under which a public function may continue to be exercised. The move is familiar to German constitutional lawyers, who would recognise it as unechte Rückwirkung: a norm attaching new consequences to a situation that began in the past but has not yet concluded. German doctrine permits this. What it does not permit is to stop there. Vertrauensschutz requires that the interference be proportionate, that a transitional period be granted where reliance interests are substantial, and that the individual situation be examined rather than presumed.

None of the three appears in the Romanian decision. Having qualified the future effects of the past situation as facta pendentia, by analogy with Article 6(6) of the Civil Code, the Court treated the constitutional question as settled by that qualification. Its reasoning, at paragraphs 121 to 123, is that the intertemporal norm imposes no personal sanction but interrupts an office whose integrity has been vitiated, and that the “progressive and profound deterioration” of the authority and image of that office is a future effect governed by the new law.

The separate opinion attacks precisely the joint at which the majority’s construction holds. Its authors argue that the unlawful conduct and the sanction attaching to it form a single normative unit which cannot be split between two successive laws, and that an administrative sanction still being served does not convert the underlying situation into a pending one. They observe that the law allows no gradual application: the thirty-day period is a dies executionis, a deadline for execution, and nothing resembling a transitional regime. And they invoke Lykourezos v. Greece, where the European Court of Human Rights held that the loss of a parliamentary mandate through an incompatibility introduced after the election frustrated the will of the electorate. That is the criticism which is both correct and useful, because it identifies exactly what a legislative correction would have to supply within a few months: a proportionality test, a transitional regime, and individual assessment.

The comparison that suggests itself is not with any of the rule-of-law files of the past decade but with the Einigungsvertrag of 1990, which allowed removal from public service on the basis of conduct predating unification and which the Federal Constitutional Court accepted only in combination with individual examination. The structure of the Romanian law is not unrecognisable to German constitutional practice. Its execution is.

The Objection That Was Made Instead

The letter argues none of this. It presents the mayor of Timișoara as a German national elected in Romania under Article 22(1) TFEU, and infers that his mandate is held by direct operation of Union law, with the consequence that its termination becomes a question of Union competence.

Article 22(1) does not say that. It provides that every citizen of the Union residing in a Member State of which he is not a national has the right to vote and to stand as a candidate at municipal elections in that State under the same conditions as its nationals. The provision governs access to the electoral process, and it does so through the principle of equal treatment. It is a prohibition of discrimination. It is not a source of the mandate, and still less a source of immunity attaching to the holder of a mandate obtained under it. Once acquired, the mandate is governed by the national law that constitutes the office.

The point becomes plain as soon as one looks at whom the Romanian law affects. It reaches more than a hundred elected officials and civil servants, almost all of them Romanian nationals. There is, on the face of it, no difference in treatment between the mobile citizen and the nationals of the host State, and therefore nothing on which Article 22 can operate. To argue the contrary is to claim that the provision confers on the Union citizen a protection that nationals do not enjoy, which turns the rule against its own rationale. No judgment of the Court of Justice supports the proposition that a local mandate is held by direct operation of Union law.

There is also a factual difficulty the letter does not address. The person concerned acquired Romanian citizenship on 18 December 2025, and Article 22 applies by its own terms to the State of which the Union citizen is not a national. But the argument fails on its own structure even without this.

Why the Argument Lands Differently Here

A legal construction is not received in a vacuum. It is received by an institutional field that already holds categories ready to absorb it, and the category it falls into is not always the one its authors had in mind.

In Romania, the proposition that a public office exercised on Romanian territory is governed directly by external law, because of the nationality of the person holding it, has a name and a history. Until the second half of the nineteenth century, foreign subjects established in the Romanian Principalities were withdrawn from the jurisdiction of local courts and answered to their consulates. They were known as sudiți, from the Italian suddito, and consular protection was routinely purchased by inhabitants who wished to escape the ordinary law. The abolition of that regime is one of the founding acts of the modern Romanian state, and it is taught as such.

Nothing in this suggests that the letter’s authors intended anything of the kind; plainly they did not. The observation is institutional rather than intentional. Arguments are absorbed by the templates available to absorb them, and the template here is old, well-established and politically active. What reads in Brussels as the effet utile of a Treaty right reads in Bucharest as the restoration of a personal jurisdiction, and it will be put to that use within days. Union citizenship was constructed as the negation of privileges of exactly that kind. Converting it into an exemption from the ordinary law of the host State hands the Romanian far right its oldest and most durable claim, which is that the foreigner is not subject to our law – and hands it over, this time, endorsed by the two largest political families in the European Parliament.

The Asymmetry

Which brings me to the comparison in the title, and it needs to be stated precisely, because it is easy to overstate.

In May 2020 the Federal Constitutional Court declared a European Central Bank programme ultra vires, held a judgment of the Court of Justice to be without binding effect in Germany, and directed the Bundesbank to withdraw from the programme unless the ECB demonstrated proportionality. Measured by any standard internal to Union law, that was an incomparably graver challenge to the Union legal order than a contested reading of the non-retroactivity clause of a national constitution.

It is not true that nothing followed. A great deal followed. The Commission opened infringement proceedings against Germany in June 2021 and closed them in December 2021, after the federal government gave formal undertakings on the primacy of Union law and the jurisdiction of the Court of Justice. The Bundestag debated the judgment and the German authorities committed themselves in writing. Every European law journal argued about it for years, from all sides, this one included.

That is the point rather than an objection to it. Every one of those responses was an instrument of law, and every one of them was deployed by an actor whose business is law: an infringement procedure opened by the Commission, a formal undertaking given by a government, parliamentary deliberation, and years of doctrinal argument. The criticism was not mild. Franz C. Mayer, writing in the German Law Journal, titled his commentary “To Boldly Go Where No Court Has Gone Before” and canvassed, among the available responses, an infringement action against his own country over a judgment of his own constitutional court. That is about as far as German constitutional scholarship goes, and it is a long way.

What did not happen is the thing that has now happened to Romania. No leader of a European political group wrote to the President of the Commission to say that Karlsruhe had been captured, that its judges were acting under political influence or had betrayed their oath, or that German money should be withheld until the judgment was reversed. The German judgment was called wrong, reckless and dangerous to the Union legal order. It was not called the output of a court that had ceased to be one, and the demand attached to the criticism was legal rather than budgetary.

The difference, then, is not one of severity. By any measure internal to Union law the German breach was the graver of the two – an express refusal of a Court of Justice judgment as against a disputed application of a national non-retroactivity clause. If severity were the criterion, the accusation of capture would have been made in Karlsruhe first, and it never was. The difference is in who speaks, in what forum, and with what instrument in hand: professors and institutions arguing about law in the German case, party leaders addressing the executive with a payment deadline in the Romanian one.

Nor does the mode of appointment supply a ground for distinguishing the two cases. The judges of the Federal Constitutional Court are elected by the Bundestag and the Bundesrat by a two-thirds majority. They are, in the plainest sense, politically appointed, exactly as their Romanian colleagues are. In the West this is described as a guarantee of pluralist legitimacy. In the East the same procedure is offered as evidence of capture.

And capture, if the word is to do any work, is an empirical claim about institutional behaviour. It can be examined. One asks how the judges are selected; whether the court’s outputs track the preferences of the alleged captor; whether dissent is possible and occurs; whether the court applies external law that constrains the domestic majority. On the first, the procedure is the German one in all essentials. On the second, in this very decision the Court annulled several provisions of the law it is accused of having endorsed. On the third, the president of the Court dissented on the very article at issue. On the fourth, on the transparency provisions the Court reached in substance the result the Court of Justice reached in C-184/20, where the Grand Chamber held that the online publication of data drawn from public officials’ declarations of interests, including data relating to a spouse or partner, goes beyond what is necessary to prevent corruption and conflicts of interest, and that data capable of indirectly revealing a person’s sexual orientation fall within the special categories of the GDPR.

On every indicator available, Decision 883 is the work of a court that applies the case law of the Court of Justice where that case law applies and reasons poorly where it does not, which is to say a court one argues with. The letter’s own evidentiary basis is the proof of it. Captured courts do not publish dissents signed by their president.

The asymmetry therefore is not a matter of tone. It determines which court is argued with and which court is reported to the Commission, and a Union that maintains it will find its rule-of-law instruments understood, in half its territory, as instruments of hierarchy rather than of law. That understanding, once established, is very difficult to dislodge, and it does not distinguish between the cases where the instruments are deserved and the cases where they are not.

The Precedent

What is being asked of the Commission is that it treat the validity of a local mandate, confirmed by a national constitutional court, as a question of Union competence, and that it act on that question through financial conditionality at the moment of maximum leverage.

Financial conditionality restores no mandate. It cannot. What it does is confirm, in the most public way available, the central proposition of the Romanian far right, which is that the country’s laws are decided elsewhere. A prime minister already removed by a censure motion and now confined to current business responded immediately by making an internal legislative correction conditional on a European verification he no longer has the standing to obtain. The law having been promulgated and the formal condition satisfied, the Commission is being asked to do the one thing it cannot do without assuming precisely the role attributed to it.

The precedent, moreover, is not the property of those who create it. A reading under which the nationality of an officeholder converts a question of national constitutional law into a question of Union competence will be available to any future majority, including one that will not resemble this one. So will the practice of settling a disputed constitutional question by declaring the court that decided it captured.

There was a way to defend the rule of law in Romania. It consisted in saying that the reasoning of the Court is weak, in saying exactly where it is weak, and in supporting those who intend to amend the law within the few months available. The other way consisted in converting a question of Romanian constitutional law into a question of Union competence on a basis that does not hold, and in attributing to a national court a condition that cannot be demonstrated. The second was chosen. It weakens the institution it claims to protect, it supplies an argument to those it claims to oppose, and it leaves behind a precedent that will outlast the case.