CETA and the Constitution of Ireland

1.1  The Tribunal administers justice, and creates a parallel jurisdiction

Article 34.1 reserves the administration of justice to courts established under the Constitution. The test is the five “characteristic features” identified by Kenny J in McDonald v Bord na gCon [1965] I.R. 217 at 231 and applied in Zalewski v Adjudication Officer [2021] IESC 24 (Dunne J at [558], [563]). A CETA Tribunal resolves a dispute as to legal rights, makes findings of breach by the Irish State on facts arising within the State, and awards damages that are then enforced against the State. Baker J accepted that this “has the potential to remove from the Irish courts the jurisdiction they currently enjoy to resolve and determine those claims”; the different law applied does not change that, because the Tribunal would “adjudicate upon state actions occurring within the State and ... make findings of breach of obligations by the Irish State arising from the same facts that might ground proceedings in the Irish courts” (at [740]). Charleton J put it more bluntly: CETA “gives a new and extra-constitutional legal mechanism” to investors aggrieved by Irish judges or Irish legislation — “effectively a parallel jurisdiction” which Article 29 does not authorise as an exercise of international relations (at [647]). Dunne J's conclusion is the holding: the creation of the Tribunal “is the creation of a parallel jurisdiction or a subtraction from the jurisdiction of the courts in this jurisdiction” (at [643](iii)).

1.2  Enforcement is “almost automatic” — the hinge of the majority

What makes the parallel jurisdiction unconstitutional rather than merely international is what happens at enforcement. The majority found, as a matter of Irish law, that an application for leave under s 25 of the 2010 Act is not a real filter. Hogan J: “it is really in the nature of a decorous formality. It certainly does not empower the High Court to review the substance of the award or anything of the kind” (at [876]). Charleton J: enforcement under s 25 “is as automatic as is possible within the Irish legal system” (at [691]). Baker J: leave “does not in itself give the Irish courts power to refuse enforcement save in the very narrow sphere of statutory defences”, interpreted in Broström Tankers AB v Factorias Vulcano SA [2004] IEHC 198 (at [745]). Dunne J identified this as the decisive distinction from the European Court of Human Rights: the ECtHR operates on the international plane and its judgments are not enforceable within the jurisdiction, whereas the “almost automatic” enforceability of CETA awards “takes a CETA award back from the international plane to the domestic” (at [610]). Under the ICSID Convention there is no public-policy ground at all (Art 54); under the New York Convention only Art V. The only refusal a court could contemplate is of the In re Greendale Developments Ltd (No 3) [2000] 2 I.R. 514 kind (Dunne J, headnote holding 4). Dunne J's holding: “it is the combination of a parallel jurisdiction together with ‘automatic enforcement’ in this jurisdiction under the provisions of the Arbitration Act 2010 that gives rise to this breach” (at [643](iv)).

1.3  Finality of Irish judgments — the concrete collision

Baker J gave the objection its sharpest form. Because an award is automatically enforceable, “the Irish High Court, Court of Appeal or Supreme Court might be compelled to enforce an arbitral award that ran directly contrary to a finding already made by a court of last resort in Ireland. It is not in my view fanciful to imagine that an investor who fails in an Irish court to obtain an award of damages on account of an unlawful state action, such as occurred in Cromane Seafoods Ltd v Minister for Agriculture [2016] IESC 6, [2017] 1 I.R. 119”, might then obtain the same damages from a CETA Tribunal and enforce them here (at [742]). Her answer to the certified question was that ratification is “incompatible with the finality of decisions of the Irish courts under Article 34” (at [792], Q2(b)). Hogan J identified the same “structural weakness” through the Vattenfall example — a tribunal expressly disagreeing with a pertinent decision of the national court — and the High Court having “no power to refuse enforcement even where the award compromised Irish constitutional identity or constitutional values in a fundamental way or where it was inconsistent with the requirements of EU law” (at [904], [808]). Art 8.22.1(f)–(g) does not answer this: the waiver bars pending domestic claims, but “the fact that a claim has concluded before a national court does not ... mean that a CETA claim may not thereafter be maintained” (Baker J at [729]).

1.6  Why the 2026 Act does not cure it — the case today

(a) A domestic refusal power contradicts the treaty, so it cannot be the cure. This was Charleton J's dissent on the remedy, and it has only gained force now that the Act exists. CETA requires the disputing party to “recognise and comply with an award without delay” (Art 8.41.2) and deems ICSID-route awards to be ICSID awards (Art 8.41.6), to which Art 54 of the ICSID Convention attaches an obligation to enforce as if a final domestic judgment, with no public-policy escape. Section 25A(3) of the 2026 Act declares such an award “not, and never was” enforceable where enforcement would compromise the constitutional order or EU-law autonomy. Charleton J: such an amendment “would, in respect of CETA Tribunal awards only, render the grounds for non-enforcement ones which contradict the terms of the treaty” (at [700]); it is “a contradiction of CETA itself; either with or without a protocol” (at [705]). Article 27 of the Vienna Convention bars reliance on internal law to justify non-performance. The Oireachtas Library's own Bill Digest recognised that the Bill “could therefore put Ireland in breach” of the Washington Convention and “could trigger a breach” under Art 27 VCLT (Bill Digest, 19 December 2025, p. 19). A safeguard the State may not lawfully use is not a safeguard.

(b) Hogan J's own warning: once ratified, EU law may forbid the refusal. The architect of the cure flagged its weakest point. Post-ratification, CETA becomes “an integral part of EU law” (Opinion 1/17, EU:C:2019:341, para 117), compliance with it becomes a “necessitated obligation” of membership under Art 29.4.6°, and “there would be at least a question mark as to whether an Irish court could properly refuse to recognise a CETA award by reference to some domestic ordre public-style considerations having regard to the duty of sincere cooperation contained in Article 4(3) TEU” (Hogan J at [879]; see also Komstroy, C-741/19, paras 23, 50). If that is right, s 25A(3)(a) — the constitutional-order limb — is precisely the power an Irish court could not exercise when it mattered. The breach the majority found was the absence of a real power to refuse; a power that EU law disables is not real.

(c) The enacted text falls short of the minimum the Court prescribed. Hogan J was “not wishing to be prescriptive”, but set a floor: the High Court must be expressly empowered to refuse where the award “materially compromised the constitutional identity of the State or fundamental principles of our constitutional order”, or “materially compromised our obligation (reflected in Article 29.4.4°) to give effect to EU law (including the Charter of Fundamental Rights and Freedoms) and to preserve its coherence and integrity” (at [1026]; [810]). Section 25A(3) instead speaks of “the constitutional order of the State” and “the autonomy of the legal order of the European Union”, drops “materially”, omits the Charter, and defines none of its terms (Bill Digest, pp. 17–18). The argument is not that the words are worse; it is that they are different, that “constitutional identity” and “constitutional order” are not synonyms, that an undefined gateway is one the High Court must construct case by case (Doyle, cited in the Digest at p. 18), and that a statute which merely defers the question cannot be assumed to meet a standard the Court set with some precision.

(d) The objections the Act does not reach. The Act addresses one thing: automatic enforcement. It does nothing about the Article 15 strict-liability objection (Hogan J), the Article 15.2 law-making objection (Charleton J), or the Article 5 Joint-Committee objection (Charleton, Baker and Hogan JJ) — the last of which no majority rejected. A challenger after ratification would run those points against a Court that has already split 3–3 on one of them.

(e) Can legislation cure a constitutional finding at all? The Court found an intrusion on the judicial power under Article 34. Article 6 vests in the People the right to designate the rulers of the State and to decide all questions of national policy, and Articles 46–47 reserve constitutional change to referendum. There is a respectable argument — raised in the Oireachtas Library's Digest (Rainford, p. 19) and not squarely decided by the Court — that where the Constitution is found to forbid something, only the People can permit it, and that six judges' assumption to the contrary was an assumption, not a holding.

(f) Sequencing. The Act commences only by Ministerial order (s 5(2)). If ratification were to precede commencement, the State would be ratifying on exactly the law Costello declared insufficient.

2.1  CETA operates on the international plane and is not law for the State

This is a majority holding, not a minority view. Art 30.6 confines CETA's rights and obligations to the Parties under public international law and bars direct invocation in domestic legal systems (at [238]). Article 29.6 of the Constitution provides that no international agreement is part of the domestic law of the State save as determined by the Oireachtas; Butler J's analysis on this point, adopting J McD v PL [2009] IESC 81 on the ECHR, was approved (at [370]–[371]). Dunne J held that although CETA creates rights and obligations for the State, “those rules are not part of the domestic legal system and could not be invoked before the Irish courts”; CETA “per se did not change the domestic law of the State” and the possibility of enforcement through the national courts “did not mean that CETA was making laws for the State” (headnote holding 2; [535], [545], [643](ii)). O'Donnell CJ ([179](2)), Baker J ([792], Q3) and Power J agreed. Crotty v An Taoiseach [1987] I.R. 713 and Pringle v Ireland [2012] IESC 47 were considered and distinguished: neither sovereignty case is engaged by an agreement that takes effect only between its Parties and by their consent.

2.2  A CETA Tribunal does not administer justice under Article 34, and subtracts nothing from the courts

O'Donnell CJ's analysis is the fullest statement of the compatibility case. The Tribunal cannot entertain any question of the validity of a measure under Irish law; a claim before it is “not a claim that was justiciable in an Irish court”; the investor “retained the same entitlement to commence proceedings in Irish courts after the execution of CETA as before”; CETA “simply gave an investor an additional option”, pursued by reference to CETA rules, normally between different parties, before a body that cannot compel witnesses and depends on a national court — which may not even be an Irish one — for enforcement (headnote; [122]). The choice “is real, but there is no sense in which it would become an irresistible option for a disappointed investor” (at [124]); the provisions are “provisions of last resort”, limited to Canadian investors, limited to compensation for actual loss, and triggered only on grounds — abusive treatment, manifest arbitrariness, targeted discrimination — “which at the level of generality would normally lead to the clear invalidity of such measures if challenged in Irish law” (at [147]). A CETA claim is “no more the administration of justice governed by Articles 34 and 37 ... than any arbitration of a cause of action in Irish law where an agreement to arbitrate was enforced by an Irish court” (headnote). MacMenamin J added the jurisdictional point: the Constitution “does not, and could not, confer on the Irish courts a jurisdiction over disputes occurring outside the State, and which do not arise under Irish law”; CETA disputes are non-justiciable here because Irish courts “would not have had jurisdiction to apply the law to which they were subject” (at [333]). Power J: an investor who elects a CETA claim “does not subtract one jot from the jurisdiction of the Irish courts”; the parallel-jurisdiction argument “conflates ... two distinctly and fundamentally different claims” (at [1060]).

2.7  The cure: prescribed by six judges, now enacted

Even on the majority's own terms the breach was conditional. Dunne J: “Were the position in relation to automatic enforcement to be altered as envisaged by Hogan J., the position would be different, and in those circumstances, CETA could be ratified without the necessity for a referendum” (at [643](vi)). Hogan J: it “would in principle be open to Oireachtas to cure the unconstitutionality identified in this judgment ... by amending the Arbitration Act 2010 in order to vest the High Court with an express jurisdiction to refuse to give effect to a decision of a CETA Tribunal” on the two stated grounds (at [810]; [1026]). Baker J agreed “with him, and my five other colleagues” that a suitable amendment “may be ... capable of providing a solution” (at [791]; [792], Q5). The three dissenters, who found no breach, necessarily agreed that an express refusal power would only strengthen the position (headnote holding 6; O'Donnell CJ at [179](6): the proposition “would only be strengthened if the 2010 Act were amended to make explicit the possibility of refusal of enforcement”).

The 2026 Act delivers each element of Hogan J's floor. It takes CETA awards out of the near-automatic ss 23–25 track altogether (s 3; s 25A(2)(ii): such an award “shall not be enforced in the State other than under this section”). It makes enforcement depend on High Court leave “subject to subsection (3)” (s 25A(2)(i)). It supplies the express refusal ground, in declaratory form, on both of Hogan J's grounds — the constitutional ground and the EU-law ground (s 25A(3)(a), (b)). And it routes any appeal directly to the Supreme Court under Art 34.5.4° (s 25A(4)), so that the court which identified the constitutional-identity concern is the court that will police it. The leave requirement is no longer a “decorous formality”; it is a statutory gateway with a declared substantive test. The one element on which the four-judge majority rested the breach — automatic enforcement — is gone.

On the drafting. The Act does replace the Court's own term, “constitutional identity”, with “constitutional order”. But the replacement has judicial pedigree. Baker J at [743] pairs exactly the Act's two concepts, “the autonomy of the EU legal system” and “the law or constitutional order of the member states”, and “constitutional order” is also O'Donnell CJ's term ([5]) and Charleton J's ([695]–[696]). “Constitutional order of the State” is, if anything, wider than “constitutional identity”, and Hogan J's own formula paired identity with “fundamental principles of our constitutional order”. “Autonomy of the legal order of the European Union” is the Court of Justice's own vocabulary from Achmea and Opinion 1/17, and is a more precise statement of the EU-law concern than a general reference to coherence and integrity. The omission of “materially” lowers, not raises, the threshold for refusal. Hogan J disclaimed prescription (“While not wishing to be prescriptive”, at [1026]); what he required was an express power to refuse on those two kinds of ground, and that is what s 25A(3) provides. Undefined terms are the ordinary material of constitutional adjudication, and the Supreme Court will define them.

On the declaratory form. Section 25A(3) does not confer a discretion. It declares, “for the avoidance of doubt”, that an award “is not, and never was, enforceable” if enforcing it would compromise either interest. That form tracks the minority's own analysis. MacMenamin J said that where a CETA decision offended “fundamental constitutional values or the constitutional identity of the State, such as judicial independence, or the finality of judgments, a court, acting under the Constitution, would have no alternative but to refuse to enforce such an award” ([342]). O'Donnell CJ agreed that a court “would be precluded from granting leave” in that case ([166]). The Act therefore codifies a limit two judges said already existed, and gives it the express statutory footing the majority required.

On the international-law objection. Charleton J's point is that the Act may place Ireland in breach of CETA and the ICSID Convention. That may be so. But the question in this note is constitutional, and a breach of a treaty is not a breach of the Constitution; on the contrary, the Supreme Court has held that the Constitution requires this safeguard. Treaty exposure is a matter for the Government under Arts 28.2 and 29.4.1°, and for negotiation (Baker J noted at [786]–[787] that the Parties were already refining the text by binding interpretation). It does not make ratification unconstitutional.

On Hogan J's sincere-cooperation worry. Section 25A(3)(b) is a power to refuse enforcement where the autonomy of EU law would be compromised — a court exercising it acts in service of Art 4(3) TEU, not against it. And the constitutional-order limb in s 25A(3)(a) corresponds to the national identity that Art 4(2) TEU obliges the Union to respect. The limbs are drafted, deliberately, in the language EU law itself recognises.

On the referendum argument. Six judges held that legislation would permit ratification “without the necessity for a referendum” (Dunne J at [643](vi)). The constitutional defect lay not in the Constitution but in a statute — the conduit through which awards acquired automatic force — and a statutory defect is cured by statute. The People's prerogative under Arts 6, 46 and 47 is engaged when the Constitution must change; here it is the Arbitration Act that has changed.