River Water Disputes between India and its Neighbours

Water is the impediment to South Asian cooperation that cannot be negotiated away, because it is the only one whose subject matter is fixed by terrain. Borders can be redrawn, migration can be regularised, insurgencies can be politically absorbed; a river’s direction of flow cannot be amended. What makes the subcontinent’s water politics distinctive is not scarcity but geometry — India sits astride almost every major basin, upstream of two neighbours and downstream of three others, and therefore has to argue two mutually inconsistent versions of international water law at the same time.

Why river water is structurally the hardest of the four impediments

• Water is a shared physical system, not a shared political problem. Cross-border migration, ethnic conflict and border disputes are all in principle severable — a fence, a citizenship register or a boundary award can end them. A river cannot be partitioned without one party losing what the other gains.
• The resource is finite in the dry season and abundant in the wet. Every South Asian dispute is a lean-season dispute; nobody quarrels over the Ganga in August. This makes the bargaining range narrow and the arithmetic zero-sum precisely when the political cost of concession is highest.
• Water disputes cross the federal boundary in a way no other impediment does.
  • Water is a State subject under Entry 17; the Union’s treaty power under Article 253 is legally sufficient but politically unusable against a State’s objection.
  • A treaty with a neighbour commits a resource that a State government administers and a State electorate consumes.
  • This is why West Bengal could block Teesta in 2011 while having no comparable purchase on, say, the maritime boundary arbitration with Bangladesh.
  • It is also why the Ganga treaty’s renewal is being negotiated with a State’s irrigation and drinking-water requirements formally on the table.
• Water is the only impediment on which India is simultaneously the accused and the complainant. Pakistan and Bangladesh accuse India of upstream behaviour; India accuses China of exactly the same behaviour, in the same vocabulary, within the same year.
• The instruments are old and the hydrology is new. The Indus allocation dates from 1960, the Ganga formula was built on flow records from 1949 to 1988, and the Kosi agreement from 1954. None of them was designed against a monsoon that is becoming more concentrated and a cryosphere that is retreating.

“The overarching Himalayas, the monsoon and the southward flowing rivers gave the subcontinent its civilizational unity; we can prosper or self-destruct together.”

Jagat S. Mehta

India’s double riparian position — the spine of the topic

The single most useful idea in this subject is that India is not a riparian of one kind. Every legal argument it makes in one direction is available to be used against it in the other, and that constraint explains its negotiating behaviour better than any account of Indian intentions.

• Upstream of Pakistan on the Indus system, and of Bangladesh on the Ganga, Teesta and Brahmaputra.
• Downstream of China on the Brahmaputra and Sutlej, of Nepal on the Kosi, Gandak, Karnali and Mahakali, and of Bhutan on the Manas, Sankosh and Wangchhu.

Basin	India’s position	The argument India makes	The argument used against India
Indus (with Pakistan)	Upper riparian	Entitlement to use its allocated share; a treaty cannot outlive the conditions that produced it	Prior appropriation and the sanctity of a binding allocation
Ganga, Teesta (with Bangladesh)	Upper riparian	Legitimate upstream needs — Kolkata port, north Bengal irrigation	Equitable and reasonable utilisation; no significant harm to the lower riparian
Brahmaputra, Sutlej (with China)	Lower riparian	Prior notification, data sharing, no significant harm, consultation on upstream works	Absolute territorial sovereignty over waters rising in Chinese territory
Kosi, Gandak, Mahakali (with Nepal)	Lower riparian	Flood moderation, storage, assured releases, navigation	Nepal’s claim to the upper riparian’s due share of benefit from its own water
Bhutan’s rivers	Lower riparian	Purchase of power, flood forecasting, joint investment	Little — this is the one relationship where the asymmetry has been made mutually profitable

• The consequence is doctrinal caution. India has never taken the Harmon position against Pakistan, because it would concede the same position to China; and it has never fully embraced the no-harm rule against China, because Bangladesh would immediately invoke it at Farakka.
• It also explains India’s preference for bilateralism. A basin-wide multilateral regime would force India to adopt one consistent doctrine across all its rivers, which is precisely what its geography makes costly.
• And it explains the asymmetry of Indian generosity. India has been most flexible where it is the lower riparian and has something to buy — Bhutanese and Nepali power — and least flexible where it is the upper riparian and has only something to give.

India argues the upper riparian’s case in the west and east and the lower riparian’s case in the north, and neither can be pressed hard without weakening the other.

The law of international watercourses

Four positions have been advanced historically; only one has survived, and it survived because the other three become unworkable the moment a state occupies more than one riparian position.

Absolute territorial sovereignty — the Harmon Doctrine

• The claim: a state may do whatever it likes with waters within its territory, regardless of downstream effect. Named for Judson Harmon, US Attorney General, whose 1895 opinion in the Rio Grande dispute with Mexico asserted the principle.

“The fundamental principle of international law is the absolute sovereignty of every nation, as against all others, within its own territory.”

Judson Harmon

• Why it is discredited. The United States abandoned it in the 1906 convention with Mexico, agreeing to deliver a fixed quantity; no tribunal has upheld it; the International Law Commission rejected it in drafting the 1997 Convention.
  • Its logical defect is that it is not a rule but a denial that any rule applies — it cannot be reciprocated, since a state asserting it upstream must accept its own destruction downstream.
• Its South Asian afterlife is real. China concedes no downstream veto over Tibetan projects, and Indian commentary during the 2025 abeyance debate reached for sovereignty language Harmonian in substance if not in name.

Absolute territorial integrity — the mirror image

• The claim: the lower riparian is entitled to the natural, uninterrupted flow of the river, and the upper riparian may do nothing to diminish it.
• Why it fails too. It freezes development permanently in the upper basin, which is why it has never been accepted either — it merely relocates the absolutism.
• Its South Asian form is the Pakistani and Bangladeshi instinct that any upstream structure is presumptively wrongful.

Limited territorial sovereignty and equitable utilisation

• This is the accepted rule. Each riparian has a right to use the waters, limited by the equal right of every other riparian — sovereignty is real but not exclusive.
• Its operative expression is equitable and reasonable utilisation, determined by weighing all relevant factors rather than by any fixed share.
  • Factors include geography and hydrology, existing and potential uses, population dependent on the watercourse, economic and social needs, effects of one state’s use on others, conservation and economy of use, and the availability of alternatives.
  • Crucially, no factor has automatic priority — not even prior use, which is why “we have always drawn this water” is an argument and not a trump.
• Community of interests is the maximal version: the basin is treated as a single unit to be developed jointly, the parties sharing benefits rather than dividing volumes.
  • The Permanent Court of International Justice in the River Oder case described a “community of interest” in a navigable river as the basis of a common legal right.
  • In South Asia it exists only in the Bhutanese hydropower model, where benefit-sharing replaced volume-sharing entirely.

Doctrine	Who it favours	Status in law	Where it surfaces in South Asia
Absolute territorial sovereignty (Harmon)	Upper riparian	Discredited; abandoned by its own author state	Chinese practice on the Yarlung Tsangpo; sovereigntist Indian commentary on the Indus
Absolute territorial integrity	Lower riparian	Never accepted	Pakistani objections to any western-river structure; Bangladeshi objection to Farakka
Limited territorial sovereignty	Balanced	The accepted rule — codified as equitable and reasonable utilisation	The reasoning of the Kishenganga awards; the Ganga treaty’s Article IX
Community of interests	Both, jointly	Aspirational; the strongest form	The India–Bhutan hydropower relationship

From the Helsinki Rules to the UN Watercourses Convention

• The Helsinki Rules on the Uses of the Waters of International Rivers (1966), adopted by the International Law Association, were the first systematic statement.
  • They introduced the international drainage basin as the unit of analysis — a hydrological rather than a political boundary.
  • Article IV entitled each basin state to a reasonable and equitable share; Article V listed the factors. Non-binding, they shaped every later instrument and were cited in the Indus arbitrations.
• The Berlin Rules (2004), also of the International Law Association, updated them — extending coverage to all waters, giving greater weight to environmental flows, groundwater and public participation, and treating the no-harm obligation more strictly.
• The UN Convention on the Law of the Non-Navigational Uses of International Watercourses (1997) is the only global treaty on the subject.
  • Adopted by the General Assembly on 21 May 1997; China, Turkey and Burundi were the only states to vote against; India abstained.
  • It entered into force only on 17 August 2014, seventeen years later, after Vietnam became the thirty-fifth party. It now has around forty parties.
  • Its core provisions: Articles 5 and 6 (equitable and reasonable utilisation), Article 7 (no significant harm), Articles 8 and 9 (cooperation and regular exchange of data), and Articles 11 to 19 (notification of planned measures and a duty to negotiate).
• Every riparian state of South Asia’s three great basins is outside it. India, Pakistan, Bangladesh, Nepal, Bhutan, Afghanistan and China are all non-parties.
  • This is the most consequential single fact about the region’s water law: there is no default regime, so where a bilateral treaty is silent or absent, nothing fills the gap.
  • China’s negative vote was not incidental: as upper riparian of the Brahmaputra, Mekong, Salween and Indus headwaters, it had most to lose from a codified duty of prior notification.

The tension the Convention did not resolve

• Equitable utilisation and no-significant-harm point in opposite directions, and the Convention deliberately left the relationship ambiguous.
  • Article 5 entitles a state to develop its share; Article 7 obliges it to avoid significant harm. A new upstream project almost always causes some harm while remaining, to the upstream state, equitable.
  • Article 7(2) softens the conflict with consultation, mitigation and, “where appropriate”, compensation — but does not say which rule prevails.
• Upper riparians read the Convention as an equitable-utilisation instrument; lower riparians read it as a no-harm instrument. That is the whole India–Pakistan and India–China argument in miniature.
  • India’s position at Kishenganga was equitable-utilisation in form: an entitlement to divert for power, subject to a floor.
  • Pakistan’s position was no-harm in form: the diversion damages an existing downstream use.
• The procedural obligations have proved more durable than the substantive ones, because notification, data exchange and consultation are where compliance can be observed — which is why India’s suspension of data sharing with Pakistan, and China’s lapsed memoranda with India, are the sharpest signals of a relationship in trouble.

The Indus with Pakistan

Partition cut the world’s largest contiguous irrigation system across a new boundary, leaving the headworks at Madhopur and Ferozepur in India and the canal commands they served in Pakistan. After the Standstill Agreement lapsed and India briefly stopped supplies in April 1948, nine years of World Bank-mediated negotiation produced not shared management but a partition of rivers — an admission that joint operation was not achievable.

The 1960 architecture

• Signed at Karachi on 19 September 1960 by Jawaharlal Nehru and Ayub Khan, with the World Bank as a signatory for specified purposes — the only third party ever admitted into an India–Pakistan instrument.
• The eastern rivers — Ravi, Beas and Sutlej — were allocated to India for unrestricted use; the western rivers — Indus, Jhelum and Chenab — to Pakistan.
  • The division is far less equal in volume than in count: roughly 20 per cent of annual flow runs in the eastern rivers, 80 per cent in the western.
  • This is the basis of the domestic criticism that Nehru conceded too much — which ignores that the eastern rivers were the ones India could physically control and that Pakistan’s canal commands were already built.
• India retains defined rights on the western rivers under Article III and Annexures C and D: domestic and non-consumptive use, limited agricultural use, and run-of-river hydroelectric generation subject to strict design criteria on pondage, spillway placement, freeboard and intake level.
• The Permanent Indus Commission under Article VIII — one commissioner from each side, meeting at least once a year, exchanging data and inspecting works. It met without interruption through three wars, which is the source of the treaty’s reputation.
• Article IX creates a graded dispute mechanism, and the grading is the treaty’s most sophisticated feature.

Stage	Instrument	What it decides	Who appoints
Question	Permanent Indus Commission	Any matter arising under the treaty	The two governments
Difference	Neutral Expert	Listed technical matters — design parameters under Annexure F, Part 1	The World Bank, if the parties cannot agree
Dispute	Court of Arbitration	Matters of interpretation or application beyond the Neutral Expert’s remit	Seven under Annexure G — two nominated by each party, three umpires by designated external authorities

• Article XII(3) and (4) are decisive for the present crisis: the treaty may be modified only by a duly ratified treaty between the two governments, and it continues in force until terminated by such a treaty. It contains no suspension clause, no withdrawal clause and no fixed term.

The two landmark outcomes before the present crisis

• Baglihar (2007) — the Neutral Expert. Pakistan objected to the 450 MW Baglihar plant on the Chenab; the World Bank appointed the Swiss engineer Raymond Lafitte in 2005 and his determination came in February 2007.
  • He upheld some Pakistani objections — reducing the permitted pondage, raising the power intake and lowering the freeboard.
  • He rejected the central one, allowing gated spillways with low-level outlets for sediment management, reasoning that treaty design criteria must be read against contemporary engineering practice and not frozen at 1960.
• Kishenganga (2013) — the Court of Arbitration. Pakistan initiated arbitration in 2010 over the diversion of the Kishenganga (Neelum) into the Jhelum basin, which affects its own downstream Neelum–Jhelum project.
  • The Partial Award of 18 February 2013 held that India may divert water for power generation, but must maintain a minimum environmental flow downstream; and separately that drawdown flushing below dead storage level is not permitted on run-of-river plants on the western rivers — a general holding of far wider effect than the plant itself.
  • The Final Award of 20 December 2013 fixed that minimum flow at nine cubic metres per second, with provision for review after seven years.
  • The Court split the difference in classic equitable-utilisation fashion: India’s right to develop, subject to a downstream floor — neither absolute sovereignty nor absolute integrity.

The parallel proceedings and India’s objection

• In 2016 both parties moved simultaneously and inconsistently. Pakistan asked the World Bank for a Court of Arbitration on Kishenganga and Ratle; India asked for a Neutral Expert on the same objections, arguing they were technical.
• The World Bank paused both processes in December 2016 to avoid contradictory outcomes, and the pause held for nearly six years while it pressed the parties to agree on one route.
• On 17 October 2022 the Bank appointed both — Michel Lino as Neutral Expert and Sean D. Murphy as Chairman of the Court of Arbitration — stating that its own powers were too limited to choose between the procedures the treaty gave each party.
• This is the procedural irregularity India objects to, and the objection is not frivolous. Article IX is a ladder — question, then difference, then dispute — and running two rungs at once risks conflicting determinations on the same design parameters.
  • India therefore refused to appear before the Court while continuing to participate in the Neutral Expert process — which is why the Court sits with five members rather than the seven Annexure G contemplates, India having appointed none.
  • The Court’s Award on Competence of 6 July 2023 held unanimously that it was properly constituted and competent, and that the parallel Neutral Expert process did not displace its jurisdiction.
  • The Neutral Expert’s decision on competence of 7 January 2025 held that the points of difference before him fell within his own technical mandate — an outcome India welcomed, since it vindicated the route India had chosen.

The notices, the abeyance and the award

• India’s notice of 25 January 2023 invoked Article XII(3) and called on Pakistan to enter negotiations to modify the treaty, citing Pakistan’s conduct on dispute settlement.
• A second notice, in September 2024, sought review and modification, broadening the grounds to fundamental changes in circumstances — demographic change, agricultural needs, the imperative of clean energy, and the effect of sustained cross-border terrorism.
• On 23 April 2025, the day after the Pahalgam attack, the Cabinet Committee on Security decided to hold the treaty “in abeyance” — a term that appears nowhere in the treaty text — with immediate effect, pending Pakistan’s credible and irrevocable renunciation of support for cross-border terrorism.
  • The formulation had been prefigured nearly a decade earlier.

“Blood and water cannot flow together.”

Narendra Modi

• Pakistan filed for review before the Court of Arbitration in March 2026. Hearings were held at the Peace Palace from 26 to 28 April 2026, with only Pakistan appearing; India did not participate.
• On 1 September 2026 the five-member Court delivered its award on the status of the treaty, holding unanimously that the Indus Waters Treaty “remains fully in force” and that India “must observe its obligations” under it.
  • The Court found no basis in international law for unilateral suspension, and rejected in turn every ground India had publicly advanced: sovereignty, Pakistan’s unwillingness to renegotiate, cross-border terrorism, demographic and energy pressures, and climate change.
  • The reasoning turns on the text itself: no suspension provision exists, and Article XII requires modification or termination by a ratified treaty between the two governments.
  • The Court also issued an order on interim measures restricting concrete work above specified levels at the Ratle dam wall and power intake, pending the Neutral Expert’s determination.
• India rejected the award within hours, restating that the Court is “illegally constituted”, that its proceedings and awards are void, and that the abeyance stands until Pakistan credibly and irrevocably ends support for cross-border terrorism.
• This was not the Court’s first ruling against the abeyance.
  • Its Supplemental Award on Competence of 27 June 2025 held that India’s declaration could not deprive it of a jurisdiction already established.
  • Its Award on Issues of General Interpretation of 8 August 2025, supplemented on maximum pondage in May 2026, construed Article III and Annexure D against India’s preferred readings of the design constraints.

What abeyance can and cannot achieve

The honest assessment resists two temptations — that India can now turn off Pakistan’s water, and that nothing has changed. What has changed is the information and inspection regime; what cannot change quickly is the volume of water crossing the boundary.

• India’s built capacity on the western rivers is a fraction of what the treaty already permitted.
  • Permitted storage is 3.6 million acre-feet; India has developed on the order of 0.2 MAF, largely the pondage at Baglihar.
  • Permitted irrigation is 1.34 million acres; roughly 642,000 acres — under half — have assured canal supply.
  • Identified hydropower potential on the western rivers is about 18,600 MW; installed capacity is roughly 3,200 MW.
• The binding constraint is engineering and geology, not law. A 300-metre Himalayan dam, its access roads and its transmission take a decade or more; the Chenab valley is seismically active and the reservoir sites are populated.
  • Even the modest Tulbul navigation barrage on the Jhelum, shelved since 1987, has gone back only as far as a project report; proposals to link the Chenab to the Ravi–Beas–Sutlej system are discussed but unsanctioned.
• What abeyance changes immediately is procedural, and procedure is not trivial.

Treaty function	Effect of abeyance	Practical consequence
Hydrological data exchange	Suspended	Pakistan loses flood and flow forecasting on the western rivers; its own reservoir operation degrades
Permanent Indus Commission meetings and tours of inspection	Suspended	No verified Pakistani sight of Indian works; disputes will be about facts, not just law
Prior notification of new projects	Withheld	Design objections can no longer be raised before construction, only after
Design constraints on pondage, spillways, freeboard	Treated by India as inoperative	Reservoir flushing and drawdown can be timed at will — small volumes, large signalling value
Actual flow volumes to Pakistan	Substantially unchanged	India lacks the storage and diversion capacity to alter annual quantum
Timing of releases in the sowing window	Marginally alterable	Short flushing pulses at Salal and Baglihar can perturb the Chenab for days, not seasons

Abeyance changes what Pakistan knows long before it changes what Pakistan gets.

• The strategic cost runs in both directions. A treaty that survived 1965, 1971 and Kargil derived its authority from being quarantined from the wider relationship; once the quarantine ends, the instrument becomes leverage rather than an agreement — and leverage is spent the moment it is used.
• The precedent problem is the sharpest cost. India’s central argument against China on the Brahmaputra is that an upper riparian owes notification, data and consultation. A unilateral suspension in the west makes that argument harder to press in the north — and Beijing has not failed to notice.
• A treaty in abeyance is not a treaty terminated. India has never claimed termination, which would engage Article XII(4) outright; abeyance occupies the space between performance and repudiation, and the 2026 award holds that no such space exists in law.

The Ganga with Bangladesh

India and Bangladesh share fifty-four rivers formally identified by the Joint Rivers Commission, established in November 1972 as one of the earliest instruments of the new relationship. That number is itself the point: no other pair of states in the region is so comprehensively tied together by hydrology, and yet only three of those fifty-four have any sharing arrangement at all.

Farakka: the engineering problem and the political one

• The Farakka Barrage on the Ganga in Murshidabad district, about eighteen kilometres from the Bangladesh border, was commissioned in 1975 after nearly two decades of construction.
  • Below Farakka the Ganga divides: the Bhagirathi–Hooghly turns south through West Bengal to Kolkata, and the Padma crosses into Bangladesh.
  • The barrage feeds a 38-kilometre feeder canal designed to divert up to 40,000 cusecs into the Hooghly.
• India’s rationale is siltation, not consumption. The Hooghly was silting and the Port of Kolkata was losing draught; a sustained headwater flush was the engineering answer proposed as early as the 1950s.
  • The claim is a non-consumptive one, which is legally the strongest form of upstream justification — the water is not used up, it is redirected.
  • Whether it has actually saved the port is contested in India too, and that argument is now made in the renewal debate.
• Bangladesh’s objection is the classic lower-riparian case. The diversion falls in the dry season, when the Padma’s flow is at its lowest and Bangladesh’s south-western districts depend on it for irrigation, navigation, groundwater recharge and — decisively — holding back the saline wedge from the Bay of Bengal.
  • The affected zone includes the Gorai offtake, the Sundarbans’ freshwater balance, and the Khulna–Jessore agricultural belt.
  • Bangladesh internationalised the dispute at the UN General Assembly in 1976 — one of very few India–neighbour disputes taken to a global forum.
• Farakka is the reason SAARC exists. Ziaur Rahman’s proposal grew out of the search for a multilateral setting in which smaller states could raise questions India insisted were bilateral — which is why the Charter ended up excluding bilateral issues.

From ad hoc arrangements to the 1996 treaty

• 1975 — a short-term arrangement for the trial operation of the feeder canal during that lean season, which India then continued without renewal.
• 1977 — the Ganga Waters Agreement, a five-year instrument.
  • It allocated 44,000 cusecs at Farakka during the driest ten-day period, of which Bangladesh received 34,500 cusecs.
  • Its distinctive feature was a guarantee clause: Bangladesh’s share could not fall below 80 per cent of the scheduled quantum in any period, whatever the actual flow. India regarded this as an open-ended liability.
• 1982 and 1985 — Memoranda of Understanding which continued sharing on similar lines without the guarantee clause, and lapsed on expiry.
• 1988 to 1996 — no arrangement at all. For eight lean seasons Bangladesh received whatever India released and flows at Hardinge Bridge fell to recorded minima. This period, not the treaty, is Bangladesh’s benchmark for what an agreement is worth.
• 12 December 1996 — the Ganga Waters Treaty, signed by H.D. Deve Gowda and Sheikh Hasina with Jyoti Basu’s concurrence as Chief Minister of West Bengal — the precedent that makes State consent a settled expectation.

The 1996 formula

• The treaty shares flows at Farakka in ten-day blocks from 1 January to 31 May, on a sliding scale keyed to the availability actually observed.

Availability at Farakka	India’s share	Bangladesh’s share
70,000 cusecs or less	50 per cent	50 per cent
70,000 to 75,000 cusecs	Balance of flow	35,000 cusecs
75,000 cusecs or more	40,000 cusecs	Balance of flow

• The design is deliberately counter-cyclical. At high flows India takes a fixed quantity and Bangladesh takes the surplus; at low flows the loss is shared equally. It is one of the more elegant allocation formulas in international practice.
• A guarantee overlays the scale: between 11 March and 10 May, the leanest window, India and Bangladesh each receive 35,000 cusecs in alternate three ten-day periods — a rotation rather than a simultaneous entitlement.
• Article II creates a Joint Committee to observe and record discharges at Farakka and at Hardinge Bridge in Bangladesh, the two gauging points on which everything turns.
• If the flow falls below 50,000 cusecs, the two governments must consult on an emergency basis and make adjustments guided by equity and no harm.
• Article IX commits both sides to conclude sharing arrangements for the other common rivers, in language taken straight from customary law.

“Guided by the principles of equity, fairness and no harm to either party, both the Governments agree to conclude water sharing Treaties/Agreements with regard to other common rivers.”

Article IX, Ganga Waters Treaty

• Article X provides for review at five-year intervals or earlier at either party’s request; Article XII fixes the term at thirty years, renewable by mutual consent.

The record, honestly stated

• The treaty ended the annual crisis, and that is not a small achievement. It removed the single largest irritant in the relationship, gave Bangladesh a predictable planning basis, and demonstrated that India would accept a binding quantified obligation to a smaller neighbour.
• Its performance has been imperfect in exactly the way its critics predicted. Studies of the record between 1997 and 2016 find Bangladesh failed to receive its scheduled share in many of the driest ten-day periods — on one widely cited count, thirty-nine of sixty.
  • The reason is structural rather than malign: the treaty allocates whatever arrives at Farakka and says nothing about augmenting it, while upstream abstraction, groundwater depletion and reduced baseflow have shrunk the pool being divided.
• There is no dispute-settlement mechanism worth the name. Unlike the Indus Treaty, the Ganga Treaty has no neutral expert, no arbitration and no third party — only consultation between two governments. Its enforcement rests entirely on political will.
• It covers one river out of fifty-four, and it stops at Farakka; it says nothing about the Ganga’s condition upstream or its delta downstream.

December 2026: renewal or lapse

• The treaty expires in December 2026 and renewal requires mutual consent under Article XII — there is no automatic rollover and no fallback regime, because neither state is party to the UN Watercourses Convention.
• The talks are live. The 86th meeting of the Joint Committee of Technical Experts met at Kolkata in March 2025 and agreed to constitute a joint technical committee on renewal; the Joint Rivers Commission met again in March 2026, and a Bangladeshi delegation has surveyed the Farakka works directly.
• India’s stated position is that any discussion proceeds through the established bilateral mechanism — the Joint Rivers Commission — and not through any special or internationalised channel.
• West Bengal’s requirements are formally in the file. The State’s drinking-water, industrial and irrigation claims, and the Kolkata port case, have been fed into the Union’s preparation. The 1996 precedent of State concurrence makes this an internal negotiation that must be concluded before the external one can be.
• The political setting has been transformed twice over.
  • Bangladesh’s government changed in August 2024 and again after the general election of February 2026; the new administration’s “Bangladesh First” framing makes water its natural first test.
  • The relationship built with one political formation now has to be re-anchored in interests, and a renewed Ganga treaty is the most conspicuous available proof that it can be.
• The substantive question is whether renewal should be replication.
  • The conservative case: renew the 1996 text unchanged, because a known formula beats a reopened negotiation that would invite Bangladeshi demands on augmentation and Indian demands on upstream use.
  • The reformist case: the formula rests on flow records from 1949 to 1988 and allocates a shrinking pool without addressing why it is shrinking; renewal should add augmentation, environmental flows, sediment management, joint basin monitoring and a dispute mechanism.
  • The augmentation question is the oldest unresolved item in the file — India has historically favoured a link from the Brahmaputra, Bangladesh a storage-based solution in Nepal, and neither has moved in fifty years.
• The cost of failure is asymmetric but shared. A lapse would return Bangladesh to the pre-1996 position, hand the water question to whoever wants to weaponise it in Dhaka’s politics, and destroy India’s best evidence that it honours quantified commitments to smaller neighbours.

Teesta and the federal constraint on Indian foreign policy

• The Teesta rises in Sikkim, runs 414 kilometres through north Bengal and enters Bangladesh at Nilphamari, where the Dalia barrage commands the Teesta Barrage Project, Bangladesh’s largest irrigation scheme. India’s Gazaldoba barrage upstream commands the West Bengal project.
• The dispute is entirely a lean-season one. Between December and April the flow can fall to a small fraction of the monsoon volume, and both barrages command more area than the river can then serve.
  • Bangladesh’s Teesta command supports rice cultivation across roughly 100,000 hectares and has repeatedly operated far below design capacity.
  • West Bengal’s case is that its own north Bengal command and drinking-water needs leave no surplus to concede.
• An ad hoc arrangement of 1983 divided the flow 39 per cent to India, 36 per cent to Bangladesh, leaving 25 per cent unallocated for the river itself — a rare early instance of an environmental flow in South Asian practice.
• The 2011 draft agreement would have given India 42.5 per cent and Bangladesh 37.5 per cent of the lean-season flow, with the balance left in the river. It was ready for signature during Manmohan Singh’s Dhaka visit in September 2011.
  • It was not signed because West Bengal withheld consent, its Chief Minister declining to travel and objecting that the allocation would leave north Bengal short.
  • Sikkim’s objections, based on its own hydropower cascade, ran in the same direction.
• This is the standing example of a federal constraint on Indian foreign policy, and it is a genuine constitutional constraint and not merely a political inconvenience.
  • The 1996 Ganga precedent, concluded with the concurrence of a Chief Minister of a rival party, established the expectation that concurrence would be sought.
  • The cost is borne in the external relationship: fifteen years of unfulfilled Indian assurances on Teesta are the most cited Bangladeshi evidence that Indian promises are not deliverable.
• The 2026 change of government in West Bengal alters the arithmetic without settling the question. The State assembly election of April 2026 produced a change of ruling party and a new Chief Minister in May 2026, removing the specific political veto that had operated since 2011.
  • It does not remove the hydrological objection: north Bengal’s command and the river’s lean-season flow are unchanged by an election, and an agreement leaving Indian farmers short will draw the same resistance under any party.
  • The realistic reading is that alignment between the Union and the State makes agreement possible rather than easy — and that a durable settlement will have to be built on augmentation and storage, not only on percentages.

The Chinese offer and Bangladesh’s use of it

• The Teesta River Comprehensive Management and Restoration Project is a Chinese proposal, developed by PowerChina, costed at roughly one billion US dollars.
  • Its components are dredging and river-regime control, embankment construction, dry-season storage, land reclamation, irrigation structures and navigation.
  • It is a management project, not a sharing agreement — it would give Bangladesh more usable water without India conceding a drop, which is exactly why it is attractive in Dhaka.
• India’s concern is location. The works sit close to the Siliguri corridor, the narrow neck connecting the North-East to the rest of India, and a sustained Chinese engineering presence there is read in New Delhi as a security matter rather than a hydrological one.
• India responded with a counter-offer during Sheikh Hasina’s June 2024 visit — Indian financing and a technical team to conduct the same restoration work — and Dhaka’s public position under both the interim and the elected governments has been that it is open to either or both.
• The episode is the clearest demonstration of hedging as a bargaining instrument. Bangladesh converted an Indian failure to deliver on sharing into leverage by inviting a third party into the same river, and India’s counter-offer is the proof that the leverage worked.

What has actually worked on the shared rivers

• The Feni agreement of October 2019 permits India to withdraw 1.82 cusecs from the Feni for drinking water at Sabroom in Tripura — a trivial volume, but the first fresh withdrawal arrangement in over two decades and a demonstration that small, non-zero-sum items can be closed.
• The Kushiyara Memorandum of September 2022 allows India to draw 153 cusecs in the lean season for the Rahimpur canal in Assam — the first water-sharing instrument between the two states since 1996.
• Flood forecasting data flows reliably. India transmits real-time flood data to Bangladesh on the Ganga, Teesta, Brahmaputra and Barak, and has extended the transmission window beyond the treaty-mandated period during severe events.
• The Joint Rivers Commission survives every political change, which is the strongest single argument that functional, technically-framed cooperation is more durable than political cooperation in this region.
• The unfinished list is long: after Teesta come the Manu, Muhuri, Khowai, Gumti, Dharla and Dudhkumar, on which interim framework agreements have been discussed for years without conclusion. Bangladesh now cites fifty-seven common rivers rather than fifty-four, and the arithmetic dispute is itself revealing.

Nepal: the upper riparian that feels like the lower one

Nepal contributes an estimated forty per cent of the Ganga’s annual flow and around seventy per cent of its dry-season flow, and holds the only viable storage sites in the entire eastern basin. It is, on paper, the strongest bargaining position of any small state in South Asian water politics. That it has produced three treaties Nepal regards as unequal and almost no built storage is the central puzzle of the relationship.

The Kosi Agreement, 1954

• Signed in April 1954 for a barrage at Bhimnagar just inside Nepal, with embankments running down into Bihar, built and paid for by India and operated by India.
  • Its object was flood control in Bihar — the Kosi is the Sorrow of Bihar for its westward migration across the plain and its catastrophic avulsions.
  • Nepal received canal irrigation and some power, but no share of benefits proportionate to its contribution of the site.
• It was revised in 1966 after sustained Nepali criticism, improving Nepal’s irrigation entitlement and its rights over the land — but the 199-year lease of the project area, written into the original agreement, survived and remains the most quoted grievance.
• The 2008 breach at Kusaha, upstream of the barrage on the Nepali side, flooded large parts of Sunsari and Saptari districts and of northern Bihar.
  • Nepal’s charge is that India, which operates the embankment, failed to take precautionary measures and then declined responsibility for damage inside Nepal.
  • It made a maintenance failure the emblematic case of Nepali powerlessness over structures on its own soil.

The Gandak Agreement, 1959

• Signed in December 1959 for a barrage at Valmikinagar on the India–Nepal boundary, again built and financed by India.
  • Nepal received canal irrigation in the Nawalparasi and Kapilvastu districts, a 15 MW power station, and a defined water entitlement.
• Revised in 1964 after Nepali protest over restrictions on Nepal’s own dry-season use and over the terms of land acquisition.
• The recurring Nepali complaint across both agreements is not that Nepal got nothing but that benefit tracked who built the structure rather than who supplied the resource — India took flood moderation and large commands in Bihar and Uttar Pradesh, Nepal a canal and a small power station.

The Mahakali Treaty, 1996, and Pancheshwar

• Signed on 12 February 1996 by Sher Bahadur Deuba and P.V. Narasimha Rao, and ratified by Nepal’s parliament that September by the two-thirds majority its constitution required for treaties on natural resources — a debate that split the Nepali left.
• It integrates three works: the colonial-era Sarada Barrage, the Tanakpur Barrage — whose afflux bund on Nepali land had been the original grievance — and the proposed Pancheshwar Multipurpose Project.
• Its central provision, and the reason Nepal signed, is the declaration that the Mahakali is a boundary river on major stretches and that both parties have an equal entitlement to its waters, without prejudice to existing consumptive uses.
  • Nepal receives defined quantities from Sarada — 1,000 cusecs in the wet season and 150 cusecs in the dry — and free power and irrigation from Tanakpur.
  • Costs and benefits of Pancheshwar are to be shared in proportion to benefits, and the treaty runs for seventy-five years.
• Pancheshwar has not been built in three decades. A 315-metre dam and about 5,040 MW across two powerhouses, it would be among the largest storage projects on earth and the only one capable of transforming dry-season flows across the eastern Ganga plain.
  • The Detailed Project Report was to be finalised within six months of the treaty. It has never been finalised.
  • The Indian consultant submitted a draft in 2016; a joint expert team has revised it since, and the first bilateral water resources talks in four years were convened in 2026.
  • The unresolved issue is the valuation of existing use. India abstracts a large annual volume through the two barrages and treats it as counting toward its post-dam entitlement; Nepal treats it as existing use to be excluded.
    • The two readings differ by billions of cubic metres, which is why the report cannot be closed.
  • Cost-sharing is the second obstacle: Nepal seeks a larger Indian share of capital cost than India has offered, and the ratio has moved without converging.
• The Nepali reading of all three treaties: India used ambiguous provisions to secure the benefits of Nepal’s geography while leaving Nepal the displacement, the submergence and none of the storage. Fair or not, it is the political fact any negotiation starts from.
• S.D. Muni describes the resulting condition as a form of denial on both sides — Nepali officials and water experts reluctant to open any new framework agreement because of what the old ones are believed to have cost, and Indian negotiators unwilling to concede that the perception itself is now the binding constraint.

Instrument	Year	What India obtained	What Nepal obtained	The Nepali grievance
Kosi Agreement	1954 (revised 1966)	Flood moderation for Bihar; large canal command	Canal irrigation; limited power; 199-year arrangement	Site on Nepali soil, benefits in India; the 2008 Kusaha breach
Gandak Agreement	1959 (revised 1964)	Barrage at Valmikinagar; irrigation in Bihar and UP	Canals; a 15 MW station	Restrictions on Nepal’s own dry-season use
Mahakali Treaty	1996	Tanakpur regularised; Sarada continued; Pancheshwar framework	Equal entitlement declared; defined releases; free power	Pancheshwar unbuilt; DPR never finalised; existing-use dispute

The counter-example: what has actually worked with Nepal

• Nepal’s theoretical hydropower potential is about 83,000 MW, of which roughly 42,000 MW is usually described as economically feasible. Its actual installed capacity is a small fraction of that, and until recently it was a net importer of electricity from India.
• Cross-border electricity trade is the one genuine breakthrough of the last decade, and it worked because it changed the subject from dividing water to selling a product.
  • Nepal began exporting surplus power to India in 2021 through the Indian energy exchange, under India’s revised cross-border trade guidelines.
  • In January 2024 the two governments signed a long-term agreement targeting 10,000 MW of Nepali electricity exports to India over ten years.
  • Approved export capacity through the Dhalkebar–Muzaffarpur and Dhalkebar–Sitamarhi lines was raised in 2026 to 1,650 MW.
  • Nepal now also exports power to Bangladesh across Indian territory under a trilateral arrangement — the first sub-regional energy transaction of its kind in South Asia.
• The lesson is transferable. Water-sharing is distributive and zero-sum; benefit-sharing through electricity is integrative, because the product did not exist before the project and both sides gain against the status quo.
• Storage remains the missing piece. Every Indian flood-control and dry-season objective in the eastern basin depends on reservoirs Nepal will not authorise until it trusts the benefit split — and every Nepali development objective depends on Indian finance and the Indian market.

Bhutan: the one model that worked, and why it has stalled

• The Bhutanese model is not a water-sharing arrangement at all. No volume is divided, no share is allocated and no gauging station adjudicates anything. India finances a project on Bhutanese rivers, Bhutan owns it, and India guarantees the market for its output.
• Four elements made it work, and each is absent from India’s other water relationships.
  • Indian capital on concessional terms — a grant-and-loan mix at a fixed rate and long tenor, so the risk sat with the financier rather than the small state.
  • A guaranteed offtake at a negotiated tariff. The output is sold into the Indian grid; Bhutan bears no market risk.
  • Bhutanese ownership of the asset after construction. Mangdechhu (720 MW), commissioned in 2019, was handed over in December 2022.
  • No downstream loss to argue about — Bhutan’s rivers are steep and short, the plants are run-of-river, and the water reaches India regardless.
• Hydropower revenue has been the largest single contributor to Bhutan’s exports and public revenue, and both sides describe the arrangement as the model for what South Asian water cooperation could be.
• Flood management cooperation runs alongside it and is genuinely joint. India funds and maintains a network of hydro-meteorological and flood-forecasting stations on Bhutanese rivers, transmitting real-time data used for flood warning in Assam and West Bengal; a Joint Group of Experts on flood management has operated since 2004.
  • The Doklam standoff of 2017 did not interrupt it — evidence against the assumption that water cooperation is always the first casualty of political friction.

The honest caveat: the model has stalled

• The 10,000 MW target set for 2020 has not been met. Four projects totalling about 2,136 MW are operational; Punatsangchhu-II (1,020 MW) was commissioned in 2025.
• Punatsangchhu-I (1,200 MW) is the cautionary case. Persistent geological instability on the right bank has caused a decade of delay and repeated redesign, and the cost overrun has been carried substantially as Bhutanese debt.
  • It is the clearest demonstration that Himalayan hydropower risk is geological before it is political, which is directly relevant to any assessment of what India could build in the Chenab valley or on the Siang.
• Debt concentration is Bhutan’s structural exposure: much of its external public debt is rupee-denominated hydropower debt serviced from a revenue stream that is itself hydropower.
  • When a project slips, the debt matures before the revenue arrives and the small economy carries the gap.
• Tariff revision is now Bhutan’s central ask. Thimphu argues that tariffs negotiated years ago do not reflect current costs or the market value of firm hydro in a grid absorbing large volumes of intermittent solar.
• Bhutan has also diversified the model, moving toward joint ventures and independent power producers with equity participation rather than the classic inter-governmental grant-and-loan structure.
• The realistic assessment: the model proves that benefit-sharing works where water-sharing cannot, but it is capital-intensive, geologically risky and dependent on a single buyer — and therefore not simply portable to Nepal at scale.

China: India as the aggrieved lower riparian

• The Yarlung Tsangpo rises in western Tibet and runs eastward for some 1,700 kilometres, turns through the Great Bend around Namcha Barwa and enters Arunachal Pradesh as the Siang, becoming the Brahmaputra in Assam and the Jamuna in Bangladesh.
  • China contributes a minority of the Brahmaputra’s total flow — most of the water is added by monsoon rainfall and tributaries below the Great Bend, in India and Bhutan.
  • But it contributes a much larger share of the lean-season flow, and it controls the sediment load and the timing, which is where the vulnerability actually lies.
• There is no treaty of any kind between India and China on any shared river. What exists is thinner by an order of magnitude than anything India has with Pakistan, Bangladesh, Nepal or Bhutan.
  • The Expert Level Mechanism, established under a joint declaration in 2006, meets annually to discuss flood-season data and emergency management. It has convened around fourteen times, most recently in 2023.
  • A memorandum on Brahmaputra flood-season hydrological data was signed in 2002, renewed at five-year intervals, and expanded in 2013 to lengthen the transmission window. It expired in 2023.
  • A parallel memorandum on the Sutlej (Langqen Zangbo) dates from 2005, prompted by the Sutlej flash flood of 2000 and the Parechu lake breach in Himachal Pradesh. It expired in 2020.
  • India pays China an annual fee for the data, which is itself an unusual arrangement and a measure of the asymmetry.
• The 2017 lapse is the case that matters. During the Doklam standoff, China did not transmit Brahmaputra flood-season data, citing damage to its collection stations, while continuing to supply data to Bangladesh under a separate arrangement.
  • Assam and Arunachal Pradesh flooded severely that season without upstream warning.
  • The episode established that China treats hydrological data as a diplomatic instrument, and that a memorandum without an enforcement mechanism provides no protection when it is most needed.
  • Transmission resumed in 2018, but both memoranda have since lapsed and renewal has been raised by India repeatedly without conclusion.

The Medog project

• In December 2024 China approved the Medog (Motuo) hydropower project on the lower Yarlung Tsangpo, and construction began on 19 July 2025.
• The scale has no precedent. Around 60,000 MW of installed capacity across a cascade of five stations, generating roughly 300 billion kilowatt-hours a year — approximately three times the Three Gorges — at a stated cost above one trillion yuan.
  • It exploits a fall of about 2,000 metres over 50 kilometres at the Great Bend, driving water through tunnels bored under Namcha Barwa rather than impounding one vast reservoir.
  • A dedicated state corporation was created to build it, and it is embedded in China’s western development and grid planning.
• India’s concerns are four, and they are not all of equal weight.
  • Flow regulation. A cascade this size can shift the seasonal distribution of releases even without net consumption — holding water in the lean season and releasing in the wet is the worst possible pattern for downstream India.
  • Sediment. The Brahmaputra carries one of the world’s heaviest silt loads, which builds the Assam floodplain. Trapping it upstream changes the river’s morphology irreversibly — the least discussed of the effects.
  • Seismicity and the outburst risk. The site sits on the eastern Himalayan syntaxis, one of the most seismically active zones on earth; the 1950 Assam earthquake originated nearby.
  • The absence of any notification obligation. Because no treaty and no convention binds China, India has no legal right to be told anything, and has had to make its case in the language of norms rather than rights.
• India’s formal response has been to request transparency, consultation with downstream states and restoration of data sharing — precisely the procedural obligations of the UN Watercourses Convention that India itself has not accepted.
• Bangladesh, the third riparian, has made the same request of China, and has an interest in India making it too — which is the strongest available argument for an India–Bangladesh common position on the Brahmaputra rather than a purely bilateral quarrel about the Teesta.

The Siang counter-move and its own contradiction

• The Siang Upper Multipurpose Project in Arunachal Pradesh is India’s answer: about 11,000 MW, a dam of roughly 280 to 300 metres, and — the point of it — some 9 billion cubic metres of storage.
  • Its stated purpose is not only power but a buffer: storage enough to absorb a sudden Chinese release and to sustain lean-season flow if upstream regulation reduces it.
  • It would also create the prior use that weighs in an equitable-utilisation assessment, strengthening India’s claim before China’s cascade is complete.
• It has been stalled for years by the communities it would displace. The Siang Indigenous Farmers’ Forum and Adi community organisations have physically prevented even the pre-feasibility survey, and the deployment of security forces to enable survey work has intensified rather than resolved the opposition.
  • The dispute is about submergence — how many villages, how much terraced land — and about a trust deficit over the project’s final footprint.
• The contradiction should be stated plainly: India objects to an upstream mega-dam built without consulting downstream populations while attempting to build one over the objections of the population immediately downstream of it.
  • This is not hypocrisy peculiar to India but the standard behaviour of a state occupying two riparian positions — though it does limit how far the normative argument can be pressed.

India’s case against China is the case Bangladesh and Pakistan make against India, and Arunachal’s case against the Siang project is the case India makes against Medog.

The frameworks the critical literature uses

Hydro-hegemony

• Zeitoun and Warner’s framework of hydro-hegemony is the standard analytical tool for asymmetric basins, and South Asia is its most obvious application outside the Jordan and Nile.
  • The argument is that outcomes in shared basins are determined less by legal entitlement than by the configuration of power among riparians — and that the observable result is usually not open conflict but a stable, unequal order.
  • Power operates in three registers: material (economic and military capacity), bargaining (setting the agenda and the terms), and ideational (making one’s own framing the accepted one).
  • Riparian position matters but does not decide: a weak upper riparian, as Nepal is, may extract less than a strong lower one.
• Compliance-producing mechanisms are the framework’s most useful contribution: hegemons secure acquiescence through coercion, utilitarian incentives, normative persuasion and treaty-based containment rather than through force.
  • The Indus Treaty is readable as containment — a settlement that removed water from the wider dispute on terms the stronger party could live with.
  • The Bhutanese arrangement is readable as utilitarian incentive — genuine mutual gain, which is why it has generated the least resentment.
  • The Nepali treaties are the case the framework handles best: formally equal instruments producing persistently unequal outcomes, and remembered as such.
• The critique is that the framework makes every outcome look like hegemony, including genuinely mutual gains, and underweights the smaller state’s agency — Bangladesh’s use of the Chinese Teesta offer being the counter-example.

Securitisation and the “water wars” thesis

• The water wars thesis holds that freshwater scarcity will be a principal cause of inter-state war in the twenty-first century, and South Asia is its favourite illustration.
• The evidence does not support it. Aaron Wolf’s survey of the historical record of transboundary basins found that cooperative interactions vastly outnumber conflictual ones, that almost all conflictual events fall short of violence, and that no modern inter-state war has been fought over water as such.
  • What water disputes reliably produce is treaties — hundreds of them — often between states hostile in every other respect. The Indus Treaty’s survival of three wars was the standard example for six decades.
• The more defensible claim is about securitisation, in the sense the Copenhagen School gives it: water becomes a security issue when a political actor successfully presents it as an existential threat, thereby licensing measures outside normal politics.
  • Pakistan’s characterisation of Indian projects as water aggression, and Indian invocations of water as an instrument of pressure after Uri and Pahalgam, are both securitising moves.
  • The 2025 abeyance is the most significant securitisation of water in South Asian history — the formal removal of a river system from technical management into the domain of national security.
  • The cost is the one the framework predicts: once water is in the security domain, the officials who can settle it are no longer the ones who understand it, and desecuritisation is far harder than securitisation.

Why bilateralism has outperformed basin management

• Every functioning water arrangement in South Asia is bilateral. The Indus, Ganga, Mahakali, Kosi, Gandak, Feni, Kushiyara and the Bhutanese projects are all two-party instruments.
• Every multilateral attempt has failed. SAARC has never taken up water, because Article X(2) of its Charter excludes bilateral and contentious issues; and there is no basin commission for the Ganga–Brahmaputra–Meghna system, the world’s largest delta.
• India’s preference for bilateralism is usually explained by power, and that explanation is partly right: in a bilateral setting the asymmetry is maximised and no coalition can form.
  • But the double riparian position explains it better. A basin regime would require India to sign up to one doctrine, and India’s interests point in two directions at once.
• The strongest case for a basin approach is technical, not political: Nepali storage benefits Bangladesh, Chinese regulation affects India and Bangladesh alike, and the Ganga’s augmentation problem has no two-party solution.
  • Sub-regional cooperation among the willing — the pattern already visible in the India–Nepal–Bangladesh power trade — is the realistic path, because it is framed around projects and products rather than shares.
• The plausible sequence is functional before political: joint monitoring and shared gauging first, then flood forecasting, then navigation and sediment management, then storage, and only then allocation.

Climate change makes every existing formula obsolete

• All the region’s allocation instruments assume a stationary hydrology. They divide a flow whose statistical properties were treated as fixed — the Ganga formula on records from 1949 to 1988, the Indus allocation on the flows of the 1950s.
• The assumption no longer holds. Himalayan and Karakoram glaciers are losing mass at differing rates, the monsoon is concentrating into intense events separated by longer dry spells, and the seasonal distribution of flow is shifting even where the annual total is not.
  • Accelerated melt yields more water in the near term, masking the problem; the long-term effect is less dry-season flow, precisely the flow every treaty divides.
  • The Indus is the most glacier-dependent of the great basins; the Ganga is the most abstraction-dependent, and each is exposed from a different direction.
• Fixed-quantity guarantees are the worst possible design under non-stationarity: a promise of 35,000 cusecs becomes harder to keep every decade, whereas a percentage share adjusts automatically.
  • The 1996 formula is partly protected by being proportional at low flows and partly exposed by its fixed guarantees in the leanest window.
• The design implication for renewal and for any new instrument is a shift from fixed volumes to adaptive rules: flow-indexed shares, explicit environmental flows, mandatory review triggers keyed to hydrological thresholds, and joint monitoring that both parties trust.

Every allocation formula in South Asia was written against a hydrological record the climate is now falsifying, and none of them contains a rule for what happens when the record stops applying.

The sources of conflict and the remedies that follow from them

Source of conflict	Where it operates	Remedy that follows
No regional legal regime — no party to the 1997 Convention	All basins	Accept the procedural obligations — notification, data exchange, consultation — even without ratifying
Structural asymmetry and the resulting trust deficit	Nepal, Bangladesh	Benefit-sharing rather than volume-sharing; visible delivery on outstanding commitments
Federal constraint on India’s treaty-making	Teesta, Ganga renewal	Institutionalise State consultation early, as in 1996, rather than seeking consent at signature
Weak dispute settlement in the eastern treaties	Ganga, Teesta	Build a neutral technical mechanism into the renewed Ganga instrument
Securitisation of water	Indus	Desecuritise: restore the technical channel first, even without political normalisation
Data scarcity and mistrust	All basins, acutely with China	Joint gauging and open hydrological data, the cheapest confidence-building measure available
Non-stationary hydrology	All basins	Adaptive, flow-indexed allocation with environmental flows and review triggers
Basin-scale problems addressed bilaterally	Ganga–Brahmaputra–Meghna	Sub-regional cooperation among the willing, starting with power trade and flood forecasting

Conclusion

River water is the impediment on which India’s structural position is genuinely difficult rather than merely inconvenient. Being upper riparian in two directions and lower riparian in three others means that every principle India invokes is a weapon it hands to someone else, and that is why Indian water diplomacy has preferred bilateral bargains to general doctrines.

• Three developments test that preference at once: an arbitral award India will not accept, a treaty with Bangladesh that expires within months, and a Chinese dam that makes India the aggrieved downstream party in the language it has spent seventy years resisting.
• The Indus shows what happens when a quarantined instrument is brought into the political relationship — the leverage is smaller than claimed, the legal exposure is larger than expected, and the precedent runs the wrong way.
• The Ganga shows that quantified obligations to a smaller neighbour are deliverable and valuable, and its renewal is the single best opportunity India has to convert goodwill in Dhaka into interest.
• Nepal and Bhutan together show the answer: where the subject is a share of water, negotiation stalls; where the subject is a jointly created benefit, it moves.
• China shows the limit of a norms-based case made by a state that has not accepted the norms, and the strongest argument for India joining the procedural regime it wants applied to others.