Before a crop is sown or a flour price is set, the water has already been divided — by a treaty, an inter-provincial accord, a federal authority, a canal statute and a village roster. This is that machinery, in order.
Ask where Pakistan's food prices are decided and most people point at a market, a mill, or a minister. The decision is older than any of those. By the time wheat reaches a shop it has passed through a chain of allocations that began at a barrage on a river, moved through a formula agreed between the provinces, and ended at a fixed turn on a village watercourse. None of that is visible at the counter. All of it is written down.
Pakistani crop agriculture is irrigated agriculture. The Pakistan Economic Survey 2024–25 calls the Indus Basin Irrigation System the world's largest contiguous irrigated network: three storage reservoirs holding roughly 13 million acre-feet (MAF) of live storage, 19 barrages, 12 inter-river link canals, some 64,000 kilometres of main canals and distributaries, and about 1.6 million kilometres of watercourses below them. Canal withdrawals in the period it covers were 60.48 MAF in Kharif 2024 and 29.43 MAF in Rabi 2024–25.
Two things follow. Water in the plains is not a natural fact but an engineered delivery, so somebody decides where it goes. And the sector it feeds is the largest block of the economy the state measures: the Survey puts agriculture at 23.5 per cent of GDP and over 37 per cent of the labour force.
Canal water is not the whole supply. The World Bank, in a March 2021 note on Indus basin groundwater, records that groundwater supplies more than half of Pakistan's agricultural water and 90 per cent of rural domestic water. Canal water arrives on a schedule and is nearly free at the point of use; groundwater must be pumped, which favours whoever can afford the tubewell.
The outer boundary was drawn internationally. The Indus Waters Treaty of 1960 allotted the eastern rivers — Ravi, Beas and Sutlej — to India and the western rivers — Indus, Jhelum and Chenab — to Pakistan. Everything below that is domestic.
The domestic answer is the Water Apportionment Accord. The Ministry of Water Resources records that it was signed among the provinces on 16 March 1991 and approved by the Council of Common Interests on 21 March 1991. Its second paragraph apportions the system's supplies: Punjab 55.94 MAF, Sindh 48.76 MAF, Khyber Pakhtunkhwa — then the North-West Frontier Province — 5.78 MAF, and Balochistan 3.87 MAF, a total of 114.35 MAF.
Three features explain most of the arguments that follow. It divides shares, not guaranteed volumes: shortages are absorbed pro rata. It settles the division between provinces, not inside them. And it left one question open — as the case study by Arif Anwar and colleagues records, the signatories accepted the need for flow below Kotri barrage to the sea but did not agree a quantity, deferring it to study.
The Accord divides a share of a variable river, not a guaranteed volume of water. In a dry year, everyone's number is smaller.
Clause 13 of the Accord called for a body to implement it. That body is the Indus River System Authority, created by the Indus River System Authority Act, 1992 — Act No. XXII of 1992, assented to on 6 December 1992. Under the Act as passed it had five members, one nominated by each provincial government and one by the federal government, with the chairmanship rotating annually. Its statutory job is to lay down the basis for regulating and distributing surface water among the provinces in accordance with the Accord, and to monitor it.
What IRSA does not do matters as much. It does not operate a canal, fix a farmer's turn, or own the water. It works at rivers, reservoirs and provincial shares; below that, distribution belongs to provincial irrigation departments. And because the Authority is a creature of ordinary legislation, its structure can be changed by ordinary legislation: an amending ordinance in February 2024 altered how it is chaired and how its committees are formed, a change the provinces did not agree on. The referee of a constitutional-scale bargain sits on a statute a single government can move.
If a province believes its share is being taken, the Constitution gives it a route, and it is not a court. Article 155 provides that where the interests of a province, or of its inhabitants, in water from a natural source of supply have been or are likely to be affected prejudicially, that government may complain in writing to the Council of Common Interests. The Council may decide the matter itself or, at a party's request, ask the President to appoint a commission, and governments must give effect to its decision faithfully according to its terms and tenor.
The Council is set up by Article 153: the Prime Minister as chairman, the four Chief Ministers, and three federal members he nominates. Article 154 requires it to be constituted within thirty days of the Prime Minister taking oath, to formulate and regulate policy on matters in Part II of the Federal Legislative List, and to meet at least once in ninety days, reporting annually to Parliament. The remedy binds; its weakness is structural, because a forum only works when it meets.
Below the distributary, water reaches a watercourse serving a block of farms and is shared by warabandi — a fixed turn. Each landholding gets a slot in a repeating cycle, timed in proportion to the area it holds. The governing statute is still the Canal and Drainage Act, 1873 (Act VIII of 1873), under which canal supplies are regulated, schedules of turns sanctioned, and interference with an outlet is an offence. Punjab layered a second structure on top with the Punjab Irrigation and Drainage Authority Act, 1997, creating the Authority, area water boards and farmer organisations.
One feature of the design decides a great deal: warabandi rations time, not volume. An entitlement is a number of minutes of whatever happens to be flowing, so a low channel means a poorer turn. And a turn at the tail of a watercourse is not the same asset as one at its head, because water is lost on the way — twice: between the barrage and the watercourse head through long, largely unlined channels, and again between the watercourse head and the field. Governments have run watercourse lining programmes precisely because those losses justify capital spending.
Warabandi gives a farmer a turn, not a quantity. Everything upstream decides what the turn is worth.
Water is delivered against land, so an entitlement is only as good as the land record. A farmer's claim rests on the record of rights: the entry showing who holds which parcel, and the mutations recording every sale, gift and inheritance since. For most of the twentieth century that record lived in handwritten registers held at village level by the patwari, who both maintained the entry and issued the copy of it — the fard — that a citizen needs for a loan, a sale or a court case. Discretion over a record plus a monopoly on the copy is a textbook concentration of leverage.
The provincial response has been to move the record into a computerised system with public service centres. Punjab did it by statute, creating a standing body under the Punjab Land Records Authority Act, 2017; other provinces have run their own computerisation programmes. What that changed is worth stating exactly: it breaks the monopoly on the copy and creates a timestamped trail, but it does not settle a disputed inheritance, move a boundary, or correct an error already in the paper register when it was scanned.
Water and land decide what a farmer can grow; two administered prices decide much of what they do grow. The first is wheat. For decades the crop was underwritten: provincial food departments and the federal agency PASSCO bought at an announced support price, provinces set a release price for flour mills, and provinces restricted the movement of grain across their boundaries under statutes such as the Punjab Foodstuffs (Control) Act, 1958. That guarantee made wheat the safe choice on a marginal acre. It began to be withdrawn in 2024: Punjab announced a support price of Rs 3,900 per 40 kilogrammes in April 2024, and reporting in October 2024 recorded a commitment under Pakistan's IMF programme that provincial governments would stop setting crop prices, with a phase-out period for Punjab.
The second is fertiliser. Urea prices turn on the price at which feedstock gas is supplied to plants — an administrative decision, not a market one — and on periodic direct subsidies; the Economic Survey for 2021–22 recorded a subsidy of Rs 1,000 per bag of DAP. Move either lever and the arithmetic of an acre changes before a seed is bought.
The retail price of flour is the last step in a chain whose first steps are a canal share, a turn on a watercourse and a procurement decision; arguing only about the last step is arguing about the smallest of them. Agriculture engages over 37 per cent of the labour force on the Survey's own count, so a water decision is an employment decision taken by people not thinking about employment. And the largest export complex, textiles, begins in a field: a short cotton crop turns a domestic water problem into a foreign exchange one.
None of this requires a new theory of Pakistan. It requires noticing that one treaty, one accord, two statutes, three constitutional articles and a village roster decide more about the price of bread than most of what is argued about on television — and that all of them are public.