Arbitration clauses are often drafted last and checked least, and the result is a genre of its own: the "pathological" clause, which names an institution that does not quite exist, designates a seat in the wrong city, or bans procedures its drafters half-remembered from another legal system. Modern arbitration law is strikingly forgiving of these defects — and strikingly unforgiving of parties who rely on them to escape arbitration. This piece works through a case study drawn from an international commercial arbitration moot problem: a single clause generating three disputes, on jurisdiction, on document production under a "no discovery" stipulation, and on the costs effect of a time-limited settlement offer.
The Clause That Got Almost Everything Wrong
The scenario: a wine buyer and a wine producer trade under a framework agreement governed by Irish law, including the CISG. A dispute over deliveries ends in arbitration, with the buyer claiming EUR 110,000 in damages plus USD 50,280 in legal costs. The agreement's dispute resolution clause reads:
"All disputes shall be settled amicably and in good faith between the parties. If no agreement can be reached the dispute shall be decided by arbitration in Dublin by the International Arbitration Tribunal (VIAC) under its International Arbitration Rules in accordance with international practice. The number of arbitrators shall be three to be appointed in accordance with those Rules. The proceedings shall be conducted in a fast and cost efficient way and the parties agree that no discovery shall be allowed. The award shall be binding and each party shall comply with the award."
The respondent seller attacks on three fronts. First, no institution called the "International Arbitration Tribunal (VIAC)" exists and no rules called the "International Arbitration Rules" exist, so the clause cannot be enforced; and VIAC, the Vienna International Arbitral Centre, sits in Vienna, not Dublin. Second, "no discovery shall be allowed" strips the tribunal of any power to order documents produced. Third, the seller has made a settlement offer (delivery of 4,500 bottles of wine to resolve the entire dispute, open for two weeks) and wants it to weigh against the buyer on costs. The analysis below draws on the Indian Arbitration and Conciliation Act 1996 alongside the UNCITRAL Model Law, the VIAC Rules 2021 and the IBA Rules on the Taking of Evidence, so its conclusions travel across Model Law jurisdictions.
Why Defective Clauses Survive: Kompetenz-Kompetenz
A pathological arbitration clause (one containing ambiguities, imprecise institutional references or procedural irregularities) is not automatically void. The controlling doctrine is kompetenz-kompetenz: the tribunal's competence to rule on its own competence. Section 16(1) of the Indian Arbitration and Conciliation Act 1996, the Indian counterpart of Article 16 of the UNCITRAL Model Law, provides:
"The Arbitral Tribunal may rule on its own jurisdiction, including ruling on any objections with respect to the existence or validity of the arbitration agreement, and for that purpose — (a) an arbitration clause which forms part of a contract shall be treated as an agreement independent of the other terms of the contract..."
Two principles are packed in here. The tribunal decides jurisdictional objections in the first instance, and the arbitration clause is separable: it survives even a finding that the main contract is null and void. Article 16 of the Model Law, adopted in over 100 jurisdictions, is to the same effect.
Courts Ask Only a Prima Facie Question
Judicial gatekeeping at the pre-reference stage is deliberately thin. In Vidya Drolia v Durga Trading Corporation, (2021) 2 SCC 1, the Supreme Court of India confined the court's role under Sections 8 and 11 of the 1996 Act to a prima facie review: disputes must be referred to arbitration unless the arbitration agreement is manifestly invalid, non-existent or incapable of being performed. Everything else (arbitrability, jurisdiction, defects in the clause itself) belongs to the tribunal under Section 16. The Court grounded this in the negative effect of kompetenz-kompetenz: the arbitral tribunal is required to look first into objections to its own jurisdiction.
Vidya Drolia built on Duro Felguera SA v Gangavaram Port Ltd, (2017) 9 SCC 729, which read the 2015 amendment to Section 11 as narrowing the court's inquiry to a single question:
"From a reading of Section 11(6A), the intention of the legislature is crystal clear i.e. the Court should and need only look into one aspect — the existence of an arbitration agreement."
Article 8 of the Model Law states the same international standard: a court must refer the parties to arbitration unless it finds the agreement "null and void, inoperative or incapable of being performed". The UNCITRAL 2012 Digest of Case Law records that this is a high threshold: mere imprecision in naming an institution or its procedural rules does not meet it, so long as the intention to arbitrate is clear and the institutional framework can be identified.
Effective Interpretation: Saving the Clause the Parties Meant
Alongside kompetenz-kompetenz runs the interpretive principle ut res magis valeat quam pereat: better that the thing have effect than be void. Academic analysis of pathological arbitration agreements, notably Dr Morten Frank's study in the Pepperdine Dispute Resolution Law Journal, describes the method courts and tribunals apply: presume the parties' intention to arbitrate where it is clearly evidenced; identify which institution the clause most plausibly refers to, even if misnamed; supplement procedural gaps by reference to that institution's rules or, where necessary, general principles of international arbitration; and leave residual jurisdictional disputes to the tribunal itself.
Applying the Framework: Wrong Name, Wrong Rules, Wrong Seat
Measured against that framework, the case-study clause holds together better than it reads.
The intention to arbitrate is unmistakable. The clause states that "the dispute shall be decided by arbitration" and that the award "shall be binding". Under the prima facie standard of Vidya Drolia, that alone establishes an arbitration agreement.
The institution is identifiable despite the misnomer. No body called the "International Arbitration Tribunal (VIAC)" exists, but the parenthetical "(VIAC)" points clearly to the Vienna International Arbitral Centre. The nomenclature is imprecise, not ambiguous as to which institution the parties intended, and effective interpretation resolves it.
The rules gap is fillable. "International Arbitration Rules" is not the formal title of VIAC's rules, which are the Rules of Arbitration (Vienna Rules). But once VIAC is identified as the intended institution, the reference is supplemented by the rules governing that institution's arbitrations, the Vienna Rules 2021. VIAC's own model clause shows how far the drafting fell short of what the institution recommends:
"All disputes or claims arising out of or in connection with this contract, including disputes relating to its validity, breach, termination or nullity, shall be finally settled under the Rules of Arbitration (Vienna Rules) of the Vienna International Arbitral Centre (VIAC) of the Austrian Federal Economic Chamber by one or three arbitrators appointed in accordance with the said Rules."
VIAC's administrative practice treats imprecise institutional references as curable by the Secretary General in consultation with the parties, with jurisdictional questions reserved to the tribunal under kompetenz-kompetenz.
The seat is the hardest defect — and still not fatal. The clause seats the arbitration in Dublin while VIAC sits in Vienna. The seat matters more than nomenclature because it determines the curial law, the law of the arbitration itself. But the error does not negate the agreement to arbitrate: the intention to arbitrate, not the precision of the seat designation, is the critical element. Once constituted, the tribunal can rule on its own seat: it may correct the seat to Vienna as VIAC's standard seat (the reading many international arbitration authorities would likely adopt once VIAC is identified as the administering institution) or honour the Dublin designation, guided by the parties' intentions and the governing law they chose. The honest caveat: a seat discrepancy is a more substantial pathology than a misnamed institution, and some courts might require a fuller conflict-of-laws analysis of the parties' likely intent before treating it as curable.
"No Discovery" Does Not Mean No Documents
Discovery and Document Production Are Different Things
The seller's second argument equates "discovery" with all document production. International arbitration practice treats them as distinct. US-style discovery is a broad, party-initiated process in which each side can demand extensive classes of documents from the other, including material not directly relevant to the issues in dispute (the "fishing expedition"). Document production in international arbitration is narrower and tribunal-managed: requests must identify documents or categories with reasonable precision, show that they are relevant to the case and material to its outcome, and survive a proportionality check on the burden of production.
The IBA Rules on the Taking of Evidence in International Arbitration (2020), widely adopted as a standard evidentiary framework even outside institutional arbitration, codify exactly this middle ground. Under Article 3, a party may submit a Request to Produce, but must describe the documents or a narrow category of documents with sufficient detail and state the reasons why they are relevant to the case and material to its outcome. Broad demands for "all documents relating to the dispute" fail those requirements by design.
The Tribunal's Default Powers and the Due Process Floor
Statutory frameworks confer evidence powers on tribunals by default. Under the Indian 1996 Act, the tribunal is the judge of the admissibility, relevance, materiality and weight of any evidence, and, failing party agreement on procedure, may conduct the arbitration in such manner as it considers appropriate, the rule Article 19 of the Model Law states for Model Law jurisdictions. The UNCITRAL framework likewise gives a tribunal a default power to require the parties to produce documents, exhibits or other evidence within a period it determines. Parties may agree to limit or exclude these powers; the point is that the law confers them initially.
But party autonomy has a floor. Section 18 of the 1996 Act requires the tribunal to treat the parties with equality and give each party a full opportunity of presenting its case; Article 18 of the Model Law is to the same effect. That right to be heard, audi alteram partem, has been understood to encompass access to material evidence and documents necessary to present one's case, a standard analysed in decisions such as Triulzi Cesare SRL v Xinyi Group (Glass) Co Ltd (2014). An agreement that eliminated document production so completely that a party could not respond to the case against it would collide with these provisions: fundamental procedural rights of this kind cannot be contracted away entirely.
What Tribunals Actually Do with "No Discovery" Clauses
Tribunals faced with "no discovery" language have routinely read it as excluding broad, US-style discovery rather than abolishing document production, and have implemented it through modified procedures: production limited to documents referenced in the parties' submissions; requests confined to specific documents or narrow categories with demonstrated relevance and materiality; no obligation to "search" for documents the producing party has not already identified or referenced; and tribunal approval of each request against necessity and burden.
On that approach, the case-study clause is a legitimate exercise of party autonomy (the parties bargained for fast, cost-efficient proceedings), but it shapes the procedure rather than deleting it. The tribunal retains the power, and the duty, to order production of documents that are relevant, material, reasonably identifiable and within the other side's possession, custody or control, because a fair hearing requires no less.
The Settlement Offer That May Carry No Costs Weight
Calderbank Offers and Section 31A
The seller's final card is its settlement offer. The device descends from the English decision in Calderbank v Calderbank [1975]: an offer made "without prejudice save as to costs" stays confidential from the decision-maker on the merits, but may be disclosed at the costs stage, where a party that rejected the offer and then did worse in the award can be penalised in costs.
Indian law now recognises the mechanism expressly. Section 31A of the 1996 Act, inserted by the 2015 amendment, gives the tribunal discretion over whether costs are payable, in what amount and when; states the general rule that the unsuccessful party pays the successful party's costs; and permits a different order, for reasons recorded in writing, having regard to all the circumstances, including, under Section 31A(3)(d), "whether any reasonable offer to settle the dispute is made by a party and refused by the other party". Indian decisions, National Highways Authority of India v Ashoka Buildcon Ltd and Oil and Natural Gas Corporation Ltd v Afcons Gunanusa JV, have been read as confirming that a Calderbank offer qualifies as such an offer.
Institutional rules point the same way. Under Article 38(2) of the Vienna Rules 2021, unless the parties agree otherwise the tribunal decides on the allocation of costs according to its own discretion, and may take into account the conduct of the parties and their representatives, "in particular their contribution to the conduct of efficient and cost-effective proceedings".
What Makes an Offer Effective
International practice and Indian jurisprudence converge on the threshold conditions. The offer must be precise enough that the tribunal can later compare it with the award and say whether the offeree would have done better by accepting. It should be unconditional, or carry only qualifications that do not undermine its genuineness. It must allow a reasonable acceptance window: English practice works from at least 21 days, and unreasonably short deadlines suggest an offer not genuinely intended to be accepted. It should be expressly marked "without prejudice save as to costs". And it should say what it does about legal costs already incurred. Offers that fail these tests lack costs effect: settlement "on terms to be negotiated", offers silent on which claims they resolve, or offers conditional on board approval or third-party consent.
The Offer in the Case Study
Measured against those requirements, the seller's offer is vulnerable on four fronts. First, quantification: 4,500 bottles is a concrete deliverable but not a monetary value, and unless the market value of the bottles can be fixed, the tribunal cannot compare the offer with the EUR 110,000 claim or with whatever it ultimately awards. Second, coverage: the buyer claims damages and USD 50,280 in legal costs, and an offer said to resolve "the entire dispute" that is silent on the costs claim is incomplete: the offeree cannot know what acceptance would extinguish. Third, genuineness: if the offer was made "without recognising any legal obligation", that caveat cuts against it, because an offeror genuinely settling must be prepared to have the offer dispose of the dispute; reserving all legal positions suggests a commercial gesture rather than a settlement. Fourth, the two-week window is very short, particularly where the arbitration is already under way and valuing the offer requires market evidence.
Even an offer that clears the threshold only feeds a discretion, not a rule. The tribunal still asks whether the offer was as good as or better than the eventual award, whether the offeree's rejection was reasonable on the information available at the time, and how both parties conducted themselves. Rejecting a vague or heavily conditioned offer is itself reasonable, and leaves the rejecting party's costs position intact.
Three caveats from the underlying research: the assessment of the offer depends on its precise wording, which the case study abstracts; whether VIAC would in fact accept administration of a case on so garbled a reference was not separately examined; and the interaction between the Irish governing law (including the CISG) and the Model Law framework applied here was not fully explored.
Practical Takeaways
- Use the institution's model clause: name the institution by its official title, adopt its rules by their formal name, and match the seat to the intended framework. The cost of a pathological clause is rarely invalidity, but it is reliably a jurisdiction fight.
- Do not bet on escaping through defects: where the intention to arbitrate is clear, courts refer the matter on a prima facie review and tribunals cure naming and rules defects through effective interpretation; a wrong seat is harder, but more likely a gap to fill than a fatal flaw.
- Draft procedural exclusions with their limits in mind: "no discovery" excludes broad, party-initiated discovery, not targeted, tribunal-managed production of relevant and material documents; the due process floor cannot be contracted away.
- Make settlement offers that can do costs work: state a monetary value or a verifiable equivalent, address every claim including costs, avoid caveats that undercut genuineness, allow a realistic acceptance period, and mark the offer "without prejudice save as to costs".
Key Authorities
- Vidya Drolia v Durga Trading Corporation, (2021) 2 SCC 1 — courts conduct only a prima facie review of the existence of an arbitration agreement; other jurisdictional objections go to the tribunal under Section 16. Source
- Duro Felguera SA v Gangavaram Port Ltd, (2017) 9 SCC 729 — under Section 11(6A) the court looks only at the existence of the arbitration agreement. Source
- Arbitration and Conciliation Act 1996, Sections 16, 18 and 31A — kompetenz-kompetenz and separability, equal treatment and full opportunity to present one's case, and the costs regime including reasonable offers to settle. Source
- UNCITRAL Model Law on International Commercial Arbitration (1985, amended 2006), Articles 8, 16, 18 and 19 — reference to arbitration, competence-competence, due process and procedural autonomy. Source
- UNCITRAL 2012 Digest of Case Law on the Model Law — "inoperative or incapable of being performed" under Article 8 is a high threshold, not met by imprecise institutional naming. Source
- IBA Rules on the Taking of Evidence in International Arbitration (2020), Articles 3 and 9 — targeted document production on demonstrated relevance and materiality; no fishing expeditions. Source
- VIAC Rules of Arbitration and Mediation 2021 (Vienna Rules), Article 38 and the Annex 1 model clause — costs allocated in the tribunal's discretion, informed by the parties' conduct. Source
- Calderbank v Calderbank [1975] — origin of offers made "without prejudice save as to costs".
- Triulzi Cesare SRL v Xinyi Group (Glass) Co Ltd (2014) — due process under Article 18 of the Model Law encompasses access to material evidence.
This analysis reflects the law as at June 2026. It is published for general information and does not constitute legal advice.