Protocol 35: Could Iceland be right after all?

1. Thesis and Standard of Review

Iceland has implemented Protocol 35 EEA in substance. The decisive question is not whether Article 3 of the Icelandic EEA Act reproduces an abstract primacy clause, but whether the domestic system gives implemented EEA rights foreseeable and judicial effect. Protocol 35 requires homogeneity through national procedures and a statutory rule of primacy only “if necessary”. The standard is therefore functional success, not the adoption of EU-law primacy or the Norwegian legislative model (Protocol 35, preamble and sole article).

The analysis has four steps: Which EEA rule was implemented? Is there an irreconcilable conflict of norms? Do interpretation, judicial protection, State liability or other mechanisms give the rule the required effect? Is there an isolated error or a structural gap that makes an additional rule necessary? ESA must identify genuine conflict cases and prove the absence of functionally equivalent instruments.

Article 3 of Act No. 2/1993 requires statutes and regulations, where appropriate, to be interpreted in conformity with the EEA Agreement and the rules based upon it. Enacted to implement Protocol 35, the provision is not a rule of invalidity, but it is a binding mechanism of interpretation and conflict resolution designed to avoid divergences and secure the practical effectiveness of EEA rights (Article 3 of Icelandic EEA Act No. 2/1993; ESA, Decision No. 212/17/COL, 13 December 2017).

An isolated error does not establish a structural deficiency. Depending on the gap, a sector-specific correction, an improved remedy or a narrowly tailored rule may suffice. The EFTA Court likewise follows an effectiveness-based standard: it derived State liability from the structure and purpose of the EEA and prohibited national rules that make damages impossible or excessively difficult (Sveinbjörnsdóttir, E-9/97; Karlsson, E-4/01). Alternative national systems of protection must be assessed by their overall effect (ESA v Norway, E-9/23).

Sveinbjörnsdóttir is particularly instructive: ESA then argued in favour of EEA State liability and thereby followed a functional approach. That position was courageous: ESA stood not only against Iceland, Norway and Sweden, but, remarkably, also against the European Commission. Because EEA law has no direct effect, defective implementation would otherwise have exposed individuals to a gap in judicial protection: they could neither enforce the EEA rule itself nor claim compensation for the resulting loss. The EFTA Court closed that gap by deriving State liability from the structure and purpose of the EEA. ESA’s former focus on practical effectiveness sits uneasily with its present formalism, which treats a particular design of primacy as decisive. Anyone now demanding a specific conflict rule must explain why the functional approach that ESA itself advocated in Sveinbjörnsdóttir should not suffice for Protocol 35.

Section 2 of the Norwegian EEA Act gives provisions serving to fulfil EEA obligations priority over other rules governing the same matter. It nevertheless presupposes a properly transformed EEA rule and a genuine conflict of norms. Where implementation is absent, late or defective, it affords no immediate protection. The Norwegian rule is therefore a limited domestic conflict rule, not a complete equivalent of EU-law primacy.

Finanger I demonstrates that limit: although the national reduction rule conflicted with the motor-vehicle liability directives, it remained applicable because implementation was inadequate. Only Finanger II opened the route to State liability (Rt. 2000 p. 1811; Rt. 2005 p. 1365). The formal primacy rule thus offered no protection in the original proceedings.

The NAV complex is still clearer: formal primacy did not prevent decades of misapplication. The administration and courts enforced residence requirements contrary to EEA law; even a criminal

conviction was initially upheld. What failed were implementation, interpretation, administrative practice, reference practice and supervision, not the existence of a conflict rule.

Further cases confirm the narrower point: a primacy rule guarantees neither a correct proportionality assessment nor effective market access, implementation of EFTA Court advisory opinions, or adequate remedies (Pedicel, E-4/04; Gaming Machines, E-1/06; Ladbrokes, E-3/06; Kristoffersen, E-8/17; Trannel, E-8/23; STX Norway Offshore, E-2/11; Netfonds, E-8/16; Fosen-Linjen, E-16/16 and E-7/18).

A functional assessment must also consider access to the EFTA Court, which private parties cannot approach directly. If national courts refuse references or State bodies press for their withdrawal, a gap in judicial protection may arise despite formal primacy (Irish Bank, E-18/11, paragraphs 53 and 57–58; Stig Arne Jonsson, E-3/12, paragraph 60; Bygg & Industri Norge, E-2/24).

Norway’s improved reference practice in recent years also qualifies the comparison. It cannot be attributed to formal primacy; in my view, it is more plausibly connected with a development in the EFTA Court’s case law that is closer to State interests and less market-oriented, making references appear less conflict-laden for State bodies. One case can therefore justify, at most, a targeted correction of the proven conflict category, not wholesale adoption of the Norwegian model.

The comparison must therefore include the institutional outlook. A functional assessment must not overlook the State and societal model prevailing in each country, the relationship between government and the judiciary, or the professional socialisation of judges—for example, the extent to which the bench is recruited from careers in the State bureaucracy. These factors may shape judicial independence, institutional reflexes and the practical willingness to enforce EEA rights against the State. A Norwegian editorial makes the concern explicit: there is broad agreement that Norway is well served by a strict immigration policy, but Norway is not well served by an expert agency that is seen to be following political signals rather than the law (Vårt Land, “Fornuftig snuoperasjon av UDI”, 11 November 2019). Nor is it well served by courts that, as law professor Terje Einarsen has observed, have in effect been instructed by the Supreme Court to trust the administration blindly (Terje Einarsen, “NAV-skandalen: Et utslag også av uvilje og tynn domstolsprøving?”, Rett24, 21 November 2019). Iceland is supported by the loyalty of its courts to the EEA judicial system: they refer questions of interpretation, take the EFTA Court’s answers seriously and secure EEA rights through interpretation or State liability. Individual erroneous decisions do not disprove that underlying approach.

Equally important is the absence of a destructive national doctrine of an autonomous “room for manoeuvre”. National discretion is legitimate only to the extent that EEA law itself permits it. Treated as an autonomous licence to dilute binding EEA obligations or neutralise unfavourable EFTA Court opinions domestically, it undermines loyalty, homogeneity and individual legal protection. Norwegian literature also warned early against detaching national discretion from the duty to perform international obligations loyally and in good faith (Mads Andenæs/Andreas Motzfeldt Kravik/Eirik Bjørge, “Høyesterett og EMD: samspill, subsidiaritet og skjønnsmargin”, Lov og Rett 2015, pp. 261–278). Formal primacy must not conceal such a doctrine.

Christoffer C. Eriksen and Ingunn Ikdahl show that this criticism is not merely external to Norway. The very title of their article, “Hemmelighold, handlingsrom og EØS”, links the discourse of room for manoeuvre with the NAV complex’s problems of transparency and control (Aftenposten, 15 November 2022). Their later work addresses the resulting requirements of good administration and the handling of border areas governed by EEA law. The necessary distinction follows: legitimate discretion exists only within EEA law; a politically or administratively asserted “room for manoeuvre” that obstructs the identification, disclosure or enforcement of binding EEA rights is not discretion, but a problem of governance and the rule of law.

2. Einarsdóttir as a Stress Test

Anna Bryndís Einarsdóttir is the stress test. In calculating her Icelandic maternity benefit, her Danish income was disregarded. The EFTA Court required a calculation based on a comparably qualified and employed person in Iceland (Anna Bryndís Einarsdóttir v Icelandic Treasury, E-5/21, judgment of 29 July 2022).

In 2024, the Supreme Court of Iceland did not set aside the administrative decision because of the unequivocal national statute, but referred to State liability. In separate proceedings in September 2026, it awarded Einarsdóttir approximately ISK 2.7 million, together with default interest and costs. She received full financial compensation, albeit secondarily through liability rather than primarily through primacy in the original proceedings (Supreme Court of Iceland, Einarsdóttir damages judgment, September 2026).

Einarsdóttir is Iceland’s Finanger: in both cases the EEA-law position did not prevail over clear national law in the original proceedings; full compensation followed only through State liability. This supports Iceland because even Norway’s express primacy rule does not ensure primary protection in every case. A possible distinction, however, supports ESA: in Finanger the directive had been inadequately implemented and section 2 was therefore inapplicable; in Einarsdóttir there may have been a genuine conflict between an already implemented EEA rule and a later national benefits rule. A primacy rule might then have operated in the original proceedings. ESA must prove that condition specifically. The case therefore justifies examination of that conflict category, not, without more, a comprehensive reordering of Iceland’s hierarchy of norms.

ESA may argue that State liability is merely secondary compensation and cannot replace the primary protection intended by Protocol 35: where an EEA rule has been properly implemented, it must prevail in a concrete conflict with later national law; Einarsdóttir could therefore reveal a structural gap in Article 3. That argument is excessive. It reduces Protocol 35 to formal principle for its own sake and gives the formal conflict rule greater weight than the system’s actual effectiveness. Protocol 35 prescribes no particular legal technique; it requires an additional statutory rule only “if necessary”. ESA must therefore show, by reference to the relevant norms and remedial practice, that there was a conflict with implemented EEA law, that Article 3 and other remedies failed structurally, and that only an abstract primacy clause could have secured effective primary protection.

Article 3 formed part of the basis of the bargain underlying EEA accession. Ratification rested on the premise that the EEA required neither a direct transfer of legislative sovereignty nor EU-law primacy. The interpretative rule was part of the compromise on which the Agreement entered into force for Iceland in 1994.

ESA’s supervisory practice is also an interpretative indicator. ESA first raised specific questions in 2012 and opened infringement proceedings in 2017. The original rule thus remained without formal objection for approximately 18 years; 23 years elapsed before proceedings were opened. This does not prove legality, but it weighs against an obvious structural deficiency existing from the outset (ESA, Decision No. 212/17/COL, 13 December 2017; ESA press release, 30 September 2026).

The timing and institutional posture also merit cautious scrutiny. It is notable that the action followed closely upon the Government’s defeat in the referendum, while statements from within the governing camp may be read as showing limited enthusiasm for mounting a vigorous defence. Any suggestion of tacit coordination would go beyond the available evidence. Even so, what appears to have been a more or less open invitation by the College to concede makes it especially important that the legal merits be tested independently, rather than inferred from the respondent State’s apparent willingness to yield.

A new general primacy rule also raises domestic questions of competence: may the ordinary legislature give broad priority to a legal order shaped externally? A parliamentary reservation permitting derogation does not answer that fundamental question (Stefán Már Stefánsson, Morgunblaðið, 10 October 2026).

3. Conclusion

Norway’s express rule strengthens protection in genuine conflicts with properly implemented EEA law. Finanger, NAV and the other case groups nevertheless show that formal primacy guarantees neither correct application nor effective individual protection. Einarsdóttir confirms this functional parallel while identifying the central issue: unlike Finanger, did it involve a genuine conflict with already implemented EEA law in which primary protection through primacy would have been possible? What remains decisive is the interaction of interpretation, reference practice, primary judicial protection, State liability and supervision, together with loyalty to the EEA judicial system and the absence of an autonomous “room for manoeuvre” that relativises binding EEA obligations.

Iceland should contest the alleged structural gap and demonstrate functional equivalence by concrete categories: remedies and outcomes, the courts’ cooperative outlook, and the absence of a destructive national doctrine of discretion. ESA must prove, in each instance, an implemented EEA rule, an irreconcilable conflict, the absence of effective judicial protection, and the necessity of an additional abstract primacy rule. In the alternative, only a narrowly tailored, constitutionally compliant amendment should be considered. The central proposition is: substance over form – ESA must not idealise Norway’s model, but must prove Iceland’s concrete functional deficiency.

Prof. Dr. iur. Dr. rer. pol. h.c. Carl Baudenbacher

Baudenbacher Law AG, Lawyers & Advisors | Door Tenant Monckton Chambers | Visiting Professor London School of Economics | President of the EFTA Court 2003-2017