Climate-proofing India’s next model BIT

India is rewriting its model bilateral investment treaty (BIT), the template for negotiating investment protection agreements. Revisions are currently underway to make it “more investor-friendly”.

The previous 2016 model BIT was deliberately restrictive, resulting in the termination of the bulk of India’s existing BITs. A parliamentary standing committee subsequently recommended revisiting the model and that revision is now underway, with public debate settled on a single question: how far should the pendulum swing back towards investor protection?

This article suggests that the pendulum is the wrong instrument. The more useful question for drafters is whether the new model is coherence-friendly: whether a treaty signed in 2026 will survive a 30-year lifespan in a legal system dense with climate obligations, to be interpreted or litigated accordingly.

Climate-proofing India’s 2026 investment treaty

India has committed itself to net zero by 2070, and to 500GW of non-fossil generating capacity by 2030. It has legislated a carbon credit trading scheme, and its exporters are being drawn into the compliance perimeter of foreign carbon border measures.

Delivering that transition requires the state to do two things that investment treaties have historically punished, i.e. to withdraw or restructure the incentives that brought renewable capital in once the technology no longer needs them; and to constrain the fossil fuel assets that earlier policy invited.

A treaty concluded in 2026 will be read for most of its working life against a climate regime that barely existed when the drafting conventions of investment treaties were settled. Unlike other fields, accelerating climate change combines the certainty of regulatory direction with the same certainty of collision, and this has already generated a body of arbitral awards running into billions of dollars. That is why climate is not one drafting consideration among many. It is the stress test that the new world has to survive.

Europe’s experience has just demonstrated at considerable cost what happens to investment treaties drafted without that acknowledgment. India now has the rare privilege of learning at someone else’s expense.

ECT’s climate clash and exits

Europe’s Energy Charter Treaty (ECT) was concluded in 1994, but its investment protections remained substantially frozen while the climate policy of its parties moved decisively. The result was litigation from both directions: fossil fuel investors suing phase-outs and renewables investors suing subsidy rollbacks, as in two dozen arbitral awards against Spain.

The end game was a visible mass withdrawal by states, with a 20-year sunset clause that kept liability running long after departure.

A modernised ECT, permitting parties to carve fossil fuel investments out of protection, finally took effect in September 2025, but this sensible amendment arrived roughly one exodus too late, and in any event does not bind departed states during their sunset periods.

The structural point for India’s drafters is twofold: a treaty that ignores the direction of international law does not escape it, but merely collides with it later at arbitral prices. Likewise, a treaty that offers its parties no means of correction short of exit will, eventually, get exits.

From exceptions to climate instructions

India’s 2016 model was an exercise in defensive engineering, and a competent one. It replaced fair and equitable treatment with a closed list of prohibited conduct. It required investors to pursue local remedies for five years before arbitrating, excluded taxation, and equipped itself with general and security exceptions.

However, the weakness of the model lies at the intersection of environment and architecture. Environmental protection appears only in the exceptions, and exceptions do their work at the end of a tribunal’s analysis, not the beginning.

A tribunal first asks whether the state’s measure breached the treatment standard. Only if it finds a breach does it ask whether an exception applies.

By that stage the state is defending itself, tribunals construe exceptions narrowly as derogations, and the burden in practice shifts to the respondent. An interpretive provision works differently; from the outset it instructs the tribunal to read the treatment standard itself in light of the parties’ environmental and climate obligations, so a good-faith regulatory measure never registers as a breach requiring excuse.

The difference is not cosmetic. It determines who bears the burden, at what stage of the case, and therefore how the dispute is priced, pleaded and settled.

The 2016 model was a defence. The 2026 model should give tribunals an instruction.

India’s BIT paths and risks

Recent Indian practice displays movement, though not unidirectional. The India-UAE BIT of 2024 retains investor-state arbitration while trimming the local-remedies period to three years and extending protection to portfolio investment.

By contrast and most prominently, the India-EFTA (European Free Trade Association) Trade and Economic Partnership Agreement, in force since October 2025, excludes investor-state arbitration altogether, pairing state-to-state mechanisms with a dedicated facilitation office and a USD100 billion investment commitment over 15 years.

These are not merely policies. They are rival theories and approaches to making foreign capital feel safe.

The UAE track says: “enforceable remedies”. An investor commits capital because it can, and in extreme circumstances hauls the state before a tribunal.

The EFTA track records institutional relationship as a case where an investor commits capital because it has privileged access, grievance channels and a counterparty with a headline number to hit, and because most regulatory disputes are better prevented than arbitrated.

Be it the UAE or EFTA track, each approach carries a distinct risk, and Europe has already demonstrated the first. Where treaty protection is fully enforceable through arbitration, a losing state often, and unpredictably, may eventually conclude that the treaty costs more than the investment it attracts, and consequently withdraw. That is the story of the ECT in a sentence.

The risk in the second EFTA approach is the mirror image. If investors hold no remedy of their own, the state’s promises are only as reliable as its desire to protect its reputation. Whether serious capital will move on that assurance alone has never been tested at scale.

India’s experiment will in time answer that question, because the EFTA agreement contains a target that can simply be checked: USD100 billion over 15 years.

The difficulty is one of timing and the answer will take a decade to emerge, while the model is currently being drafted. While the drafters can choose either dispute-settlement design, the substantive provisions proposed below are fully functional under either design.

Where there is a tribunal, they instruct it; where there is only state-to-state consultation, they supply the standards against which that consultation is conducted. The choice of dispute mechanism is a wager on an open question. The substantive text is the part of the treaty that wins either way.

Mapped models for India’s BIT

None of this territory is unmapped. Other countries have already experimented with different treaty designs and India’s government can learn from them.

Brazil has run India’s second track since 2015: its Cooperation and Facilitation Investment Agreement model dispenses with investor-state arbitration entirely, relying on an ombudsperson, joint committees and dispute prevention, with state-to-state arbitration as the backstop. Under this, Brazil has signed more than 30 such agreements while continuing to attract substantial foreign investment.

India knows this model from the inside, having signed a treaty on precisely that pattern with Brazil in January 2020, five years before the EFTA agreement made the approach look novel.

At the other end of the spectrum, the Nigeria-Morocco BIT of 2016 pioneered a different kind of coherence and, while not yet in force, the treaty in principle imposes obligations on investors’ environmental and social impact assessment, anti-corruption compliance, and labour and human rights protections.

Arbitral practice has begun to move the same way. Tribunals have accepted jurisdiction over state counterclaims. Ecuador recovered USD41 million from an investor for environmental damage. A treaty that obliges investors as well as states is not a curiosity, it is where the field is heading.

For a negotiator, however, the most consequential fact is nearer home. India has been negotiating an investment protection agreement with the EU since 2022, alongside its agreements with the UK.

The EU’s own post-CETA (Comprehensive Economic Trade Agreement) treaty practice already insists on an operative right to regulate and express references to the Paris Agreement. The Netherlands model of 2019 does the same.

In other words, coherence provisions proposed below are not concessions India would be making to investors or to counterparties. They are, in substantial part, the very provisions its largest pending counterparties will seek. A drafter converts such anticipated demand into an opening position – and an opening position into negotiating capital – while building the model. It is not often that sound policy and cheap diplomacy arrive in the same clause.

Six low-cost climate-coherent clauses

Whatever position the new model takes, six drafting choices would cost little and prevent much. These would avoid the pitfalls of earlier treaties and ensure climate-coherence.

(1) An express systemic integration clause. Article 31(3)(c) of the Vienna Convention already directs tribunals to consider other rules of international law binding between the parties, but arbitral engagement with the provision is sporadic and unpredictable.

The model should remove the discretion with a provision along these lines: “In interpreting this treaty, a tribunal shall take into account, together with the context, the parties’ obligations under other international agreements to which both are party, including agreements relating to climate change and the environment.”

(2) An operative right to regulate. The 2016 model’s concern for regulatory autonomy lives in its preamble and exceptions. The new text should follow the technique of the EU-Canada agreement, which states that the mere fact of regulating in public interest, including protecting the environment, does not of itself constitute a breach.

(3) A calibrated climate carve-out rather than a self-judging one. A self-judging clause protects any measure the state itself “considers necessary” for climate purposes. The state’s own say-so is conclusive and no tribunal reviews it. The temptation, after watching Europe, is to draft exactly that.

But it should be resisted, for two reasons. As a matter of negotiation, no capital-exporting counterparty will accept unreviewable immunity, so the clause becomes a bargaining liability.

In India’s case domestic companies are themselves substantial outbound investors in renewables and infrastructure, which means a clause India drafts as a shield will be quoted back at Indian claimants abroad.

The workable middle is an objective test. A measure is protected if it is adopted in good faith for climate mitigation or adaptation, applied without discrimination between comparable investors, and not manifestly disproportionate to its stated objective.

That formulation protects genuine climate regulation while leaving disguised protectionism exposed – which is precisely the line a well-advised investor also wants drawn.

One point of placement follows from the earlier discussion. This provision belongs in the treatment standards themselves, phrased as “shall not constitute a breach”, and not in the exceptions chapter where, as demonstrated, it would arrive too late.

(4) Define legitimate expectations. Under the fair and equitable treatment standard, tribunals protect an investor’s “legitimate expectations”. But treaties never say where such expectations may come from and arbitral practice has been split on the question. On the broad view, the regulatory framework in force is itself an implicit promise: change it to the investor’s detriment and compensation follows.

On the narrow view, only a specific commitment addressed to the investor will do.

The new model should close the question the ECT left open, with a provision to this effect: “Legitimate expectations arise only from specific written commitments in a contract, a statutory guarantee addressed to a defined class of investments, or a targeted representation by a competent authority on which the investor demonstrably relied at the time of investment.”

General legislation and statements of policy do not qualify. The practical effect is best seen from the investor’s side of the table.

A solar developer entering India under such a treaty knows precisely what paper to obtain before committing capital – a tariff in a power purchase agreement, or a statutory rate addressed to solar generators – and knows equally that ministerial speeches count for nothing in a later dispute.

The state, for its part, knows exactly which undertakings it may not disown. This is conventionally described as a pro-state provision. It is better described as a rule that tells both sides before the money moves.

(5) A joint interpretive committee with binding authority. The ECT’s deepest design flaw was not any single protection but the absence of a steering mechanism. Between acquiescing in whatever tribunals decided and withdrawing altogether, its parties had nothing. Yet the precedent for the intermediate gear has existed for decades.

Under the North American Free Trade Agreement (NAFTA), the treaty parties could issue joint interpretations binding on tribunals, and used the power in 2001 to rein in an expanding reading of fair and equitable treatment. The EU-Canada agreement carries the same mechanism.

The new model should establish a standing India counterparty committee empowered to adopt interpretations that bind tribunals, prospectively. If arbitral practice under the treaty begins to drift, as it did under the ECT, the parties can correct course. A treaty, like any long-term contract, is safer with a variation clause than with only a termination clause.

(6) The sunset. Survival clauses – provisions protecting existing investments after termination – commonly ran 15 or even 20 years in the ECT era, which is why Europe will be arbitrating with departed states well into the 2040s.

India is living the same lesson domestically: the BITs it terminated after 2016 continue to support claims under their own survival clauses. Given that experience on both continents, no one drafting in 2026 should copy the 20-year ghost.

A five-year survival period preserves the legitimate reliance of existing investors while ensuring that a failed treaty can be allowed to expire. Drafting one’s exit terms while relations are friendly is not pessimism; it is the same prudence the treaty demands of investors.

Model treaty faces legacy leaks

No matter how well drafted, a model treaty does not enter an empty field. When India terminated its BITs after 2016, the investment chapters embedded in its trade agreements survived. Agreements with Singapore (2005), South Korea (2009) and Japan (2011) all contain older-style protections, including fair and equitable treatment and investor-state arbitration, and none can be unilaterally terminated without exiting the trade agreement.

The consequence deserves plain statement. Whatever discipline the new model imposes, a Japanese, Korean or Singaporean investor already holds the undisciplined version, and an investor of any other nationality can acquire it by structuring its investment through Singapore, for example.

The new model will govern the treaties India signs next. It does nothing to the treaties India could not unseal. Until the legacy chapters are renegotiated, the new architecture leaks – and that leak becomes a flood if the revised model reintroduces a most-favoured-nation (MFN) clause.

MFN is the master key of investment treaty law. It permits an investor under one treaty to borrow more generous provisions from another. Under “investor-friendly” pressure, the temptation to restore MFN will be real – and yielding to it could undo all the careful drafting of climate coherent provisions.

This suggests a seventh provision, homegrown rather than imported: omit MFN, as the 2016 model did, or confine it expressly so that it cannot import substantive standards or dispute settlement provisions from other treaties. Of the 2016 model’s defensive choices, this is one revision that must not be softened.

Trade law imposes a second discipline. A measure sheltered by the new model’s climate carve-out is not thereby immune from India’s trade obligations, a lesson India has already been taught.

In 2016, the World Trade Organisation ruled in the India-Solar Cells dispute with the US that domestic content requirements of the National Solar Mission, a flagship climate programme, violated the national treatment obligation. The same measure can thus be safe under the investment treaty but condemned under trade disciplines, with the state reading two inconsistent verdicts on one regulation.

The drafting response is alignment. The climate carve-out in the new model should track the general exceptions language of India’s trade agreements so that the two regimes test a measure against the same words. Otherwise, fragmentation, shown out through the front door, returns through the back – an awkward outcome for a model that has a premise of coherence.

Institutional overlap requires the same attention, and the UAE relationship is the test case. A UAE investor’s substantive protections live in the BIT of 2024; the CEPA of 2022 governs trade and services, and speaks mainly state to state. The two instruments nonetheless meet on the same ground because a single Indian measure, a licensing condition or a change to energy tariffs may simultaneously engage the CEPA commitments and BIT’s investment protections.

Managing BIT-CEPA friction and overlap

Three frictions follow: (1) the same measure may be tested against different standards and graded differently by each instrument; (2) proceedings may collide, the investor arbitrating under the BIT while its home state raises the same measure through the CEPA’s state-to-state machinery, with no rule co-ordinating the two or preventing inconsistent findings; and (3) interpretations may diverge.

The joint committee proposed above derives its authority from the BIT and binds only BIT tribunals, while the CEPA’s committee answers to its own text, and neither can instruct the other. For future treaties, the drafting responses are standard. A conflict clause provides that, in the event of inconsistency, the investment treaty prevails to the extent of the inconsistency. An election-and-waiver provision requires an investor commencing arbitration to abandon other avenues with respect of the same measure, matched by an undertaking of the parties to suspend duplicative state-to-state proceedings.

Lastly, the incorporation of a consultation duty, written into both instruments in mirrored terms, requires each committee to confer with the other before interpreting language the two instruments share.

Three vehicles to update treaties

But the UAE pair is already signed, which raises a practical question: how does any of this reach a treaty in force?

The solution is by three vehicles, in ascending order of formality. The first is an administrative arrangement between the two governments. Co-ordination between committees needs no amendment at all because the committees are the parties’ own organs, and the parties may simply direct them to confer.

The second is a joint interpretive agreement, which article 31(3)(a) of the Vienna Convention obliges tribunals to take into account. It serves to clarify what existing provisions mean, and India has used precisely this device before, issuing Joint Interpretative Notes with Bangladesh for their BIT in 2017.

The third is a protocol of amendment, required for anything that cuts into an investor’s procedural rights, above all, the election-and-waiver rule. Such a protocol should apply prospectively to claims commenced after its entry into force. The allocation matters because the vehicles are not interchangeable: a purported interpretation that functions as an amendment invites tribunals to resist it, as the NAFTA parties briefly discovered in 2001, before their interpretive note prevailed.

The rule of thumb is candour. Interpret what is genuinely ambiguous; amend what is genuinely new.

One proposal, finally, eases friction rather than adding to it, and the mechanics show why. The systemic integration clause and the MFN clause both look outward to other treaties, but they act on different objects. Systemic integration acts on meaning.

When a tribunal asks what this treaty’s words require, whether a measure was adopted in good faith or whether treatment was fair, the clause directs it to answer with the parties’ other agreements in view.

The other agreement never becomes applicable law; no new obligation enters the treaty; the investor cannot sue for its breach. The words being interpreted set the outer limit because interpretation can do no more than the text will bear.

MFN adds content beyond text

MFN, by contrast, acts on content. It entitles the investor to treatment no less favourable than that promised to investors of third states, which permits a tribunal to lift a more generous provision out of another treaty and apply it as though it were integrated into this one.

That is not the interpretation of existing text but addition of new text, and its only limit is what India’s other treaties happen to contain. Hence the asymmetry in this article’s recommendations: the new model can safely direct tribunals to read it alongside India’s trade agreements, but it cannot safely leave an unconfined MFN clause near them.

One caveat keeps the claim honest. Reading two instruments together will not reconcile texts that genuinely diverge; that is the work of the above-mentioned alignment exercise. What it prevents is each regime deciding in ignorance of the other.

Coherence, if it is to mean anything, must run horizontally across India’s own treaty network as well as vertically against climate law, and the drafters should test the new model against both axes before either is litigated.

Interim safeguards before new treaties

The revised model is not yet public and treaties negotiated from it are years from being in force.

In the interim, investors structuring into or out of India should not assume treaty protection will resemble the pre-2016 vintage. Protection should be built at the level of the investment itself, through contractual stabilisation clauses, specific written assurances from the competent authority and, where available, structuring through the India-UAE BIT’s fuller protections.

Counsel advising states or advising against them should plead the parties’ climate obligations expressly and early, under article 31(3)(c), rather than as atmospheric background in the closing submissions. The interpretive future described above can, to a substantial degree, be pleaded into existence now.

The provisions proposed here are not exotic. Each has a working precedent in Brussels, Ottawa, The Hague, Brasília, Abuja or, in the case of the last, New Delhi itself. What no major state has yet done is assemble them into a single unified model designed for the climate era from the outset.

The 2016 model answered the question its decade asked: how does a state stop losing arbitrations it did not expect? The question this decade asks is how a treaty remains stable while the law around it accelerates?

India holds an unusual hand: a revision already underway, live negotiations with counterparties who will demand half of these provisions anyway, and a decade of comparative experience both abroad and at home to draw on.

The opportunity is presented to not merely update the template but produce the first major model treaty of the post-ECT era – for once, an Indian legal export that arrives ahead of demand.

---

Mikhail Behl is an arbitrator, advocate (India) and barrister (England and Wales). He has held more than 125 arbitration appointments including in state party disputes. Behl is empanelled with leading arbitration institutions including the Singapore International Arbitration Centre, Hong Kong International Arbitration Centre and World Intellectual Property Organisation Arbitration and Mediation Centre.

 

---