CESR Global Partnerships Lead Mahinour ElBadrawi along with civil society allies pictured at the negotiations.
From principles to plumbing: what the fifth session of the UN tax negotiations revealed
By Juan Auz, Fiscal Justice Lead
Between 3 and 13 August, governments met in New York for the fifth session of the Intergovernmental Negotiating Committee drafting the UN Framework Convention on International Tax Cooperation. For the first time, they worked from a full draft treaty: twenty-six articles released only on 21 July, alongside drafts of the two early protocols. Earlier rounds asked whether the world would rewrite tax rules at the UN, whereas this one asked what those rules will say. As the CESR's daily updates from the room have shown, the answer is being decided in the nitty-gritty of certain provisions.
What the argument is actually about
Strip away the article numbers and one question sits underneath everything: when a company makes money in a country, which government gets to tax it? Rules written a century ago treat a multinational as a bundle of separate firms trading with each other, so profit can be booked wherever the group claims the value was created. This value creation frequently occurs in a jurisdiction with a very low tax rate where it has few employees or assets. Article 5 of the draft Convention would start taxing companies where they actually operate: where the staff, customers, and sales are.
The amounts at stake are not marginal. Research by Public Services International and the Tax Justice Network, released as the session opened, estimates that honoring that commitment would raise roughly US$500 billion a year in additional corporate tax worldwide without any government raising a single rate. For many lower-income countries, it would be several times what they currently collect from multinationals: money to implement fundamental rights by paying for nurses, classrooms, care systems, and climate adaptation.
Ambition, in other words, can be made or diluted in the fine print. The fifth session is where this Convention's fine print began to be written.
Week one: the Convention's load-bearing walls
Three takeaways stand out from the first week.
First, the argument has shifted from whether to how much. Almost no one now contests that the Convention will exist. Day one was spent defending Articles 1 and 2 – the objectives and guiding principles, including the commitment to align tax cooperation with States' human rights obligations – against attempts to reopen what the General Assembly settled in the 2024 Terms of Reference, which Nigeria called the constitution of this process. Yet, Norway asked whether “fairness” carries any legal meaning; and Germany argued that the instrument should contain principles rather than binding obligations.
Second, dilution is subtly instilled through procedure rather than open opposition. Some States, especially those pertaining to or aligned with the OECD, that endorse the Convention's aims have nevertheless pressed for a standalone sovereignty article, reservations (declarations by which a State excludes itself from or modifies specific provisions), consensus decision-making, a facilitative rather than governing Conference of the Parties, and voluntary rather than assessed financing. Each proposal might be defensible in isolation; however, together they describe an instrument that commits to a great deal and requires very little. The Secretariat responded during the heated debates, offering guidance, asserting that where a treaty is silent on decision-making, established practice is consensus. But silence is of course a choice.
Third, the Convention's ambition is now concentrated in three provisions: articles 5, 13 and 21. Article 5 on the fair allocation of taxing rights over multinationals – which the African Union called the reason delegations came to New York – has lost “economic activity” from its list of nexus factors (nexus being the connection that entitles a country to tax an activity at all). Economic activity now joins the remaining factors with “and” rather than “or”, making them cumulative. Article 13 determines whether the Conference of the Parties can steer implementation or merely convene. Article 21 governs the relationship with the thousands of existing bilateral tax treaties, and drew the sharpest drafting point of the session from Zambia: a duty to align treaties that bites only when another party requests renegotiation is an obligation deferred. Civil society added the corollary – under Article 30 of the Vienna Convention on the Law of Treaties, the later treaty would ordinarily prevail. Hence, the old network governs only because that option might end up in the Convention.
Two further matters of inequality are significant Taxation of high-net-worth individuals remains in the Convention rather than in either protocol. Yet, the draft softened from “develop and implement” to “cooperate to enhance”, and Mexico noted that it treats the very wealthy solely as a risk of evasion rather than asking whether they pay their share. Separately, a bloc of European states, along with Japan, Korea, and Singapore, wants tax avoidance removed from the definition of illicit financial flows because “avoidance is lawful”. Nigeria reacted by arguing that the largest losses to developing-country treasuries stem from arrangements that are lawful at every step.
Week two: the protocols, and the sequencing problem
The second week turned to Protocol 1 on taxing income from cross-border services and Protocol 2 on preventing and resolving tax disputes. Negotiating protocols for an unfinished Convention creates an obvious difficulty: objectives, scope, definitions, and the treatment of existing agreements are hard to settle before the parent commitments are in place. Protocol 1's relationship with prior treaties was expressly deferred until Article 21 is resolved.
But the connection runs deeper than sequencing. Protocol 1 is where Article 5 (allocation of taxing rights) either becomes operational or does not. If its rules continue to rest on physical presence, and if maximum source tax rates leave countries worse off than under their current treaties, the Convention's redistributive promise evaporates in the drafting. The dispute over gross-basis withholding (taxing the payment) versus net-basis taxation (taxing profit after deductions) is not a technical preference but a redistributive question since gross taxation is simpler to administer and harder to manipulate, which is why Global South administrations favor it.
Protocol 2 poses the mirror risk. Building dispute prevention around transfer pricing and the arm's length principle – the rule that treats affiliates of one multinational as if they traded at market prices, and the main channel through which profits are shifted – would embed the system the Convention was meant to reform. Advance pricing arrangements and taxpayer-initiated procedures are costly for tax administrations and advantageous to those who can afford advisers, and making arbitration optional resolves none of the concerns about cost, constitutional authority or the pressure powerful treaty partners can exert. Beneath both texts lies a quieter question CESR has pressed throughout: whether multinationals are being written in as rights holders entitled to procedural protection rather than as duty bearers owing a do-no-harm obligation. Both protocols are already optional to ratify; the further fight over opting in and out of individual provisions proposed by some countries would hollow out whatever common core survives.
Familiar patterns from other negotiations
Delegates borrowed openly from other regimes. Belgium, for example, proposed that the sovereignty article be modeled on the UN Convention against Corruption, which also contains, in Art. 63, a near-universal call for observer provisions; and Côte d'Ivoire observed that the financing provision tracks the tobacco convention's reliance on unpredictable, voluntary funding. The comparisons cut both ways: framework-plus-protocol architectures also allow ambition to be deferred into instruments few ratify.
If we look at other multilateral negotiation processes for comparison, the binding treaty on business and human rights has held 11 sessions since 2015 without an agreed text, largely because the home states of the corporations concerned disengaged. The plastics negotiations have collapsed twice over the consensus rule, prompting researchers to call in Nature for majority-fallback voting when a minority blocks broad support, precisely the question raised by Article 13. The tax process has so far avoided both fates: it has a General Assembly mandate, Terms of Reference adopted by vote and a fixed 2027 deadline, and it has proceeded despite the United States' withdrawal from the talks. Its risk is not collapse but attrition and lack of ambition. The Agreement on Marine Biological Diversity Beyond National Jurisdiction, in force since January 2026, shows that a Global South–led package can succeed, albeit after nearly twenty years of drafting and negotiations and amid a different geopolitical convergence of interests.
What comes next
Written submissions are due in late August, and revised texts are expected before the next session, to be held in Nairobi. As CESR argued after Nairobi last year, success is not measured by whether a Convention and two protocols are produced, but by whether they enlarge the resources States can mobilize for health, education, care, gender justice and climate action; i.e. the obligation, under international human rights law, to devote the maximum available resources to realizing rights.