Covering Pakistan, Bangladesh, Sri Lanka, Nepal, Bhutan, the Maldives and Afghanistan.
The protein system
This is a region of 536 million people[72] in which Pakistan and Bangladesh together account for 81 per cent of consumption, so the regional score is largely their score, adjusted for an Afghan humanitarian collapse and three small import-fed states.
Protein supply per head is adequate on paper and thin in composition, running from 62.58 grams per day in Afghanistan to 81.74 in Nepal and 97.41 in the Maldives, with Pakistan at 73.39 and Bangladesh at 67.99.[1] Pakistan draws 32.55 grams of animal protein per person per day and the Maldives 53.03, against 8.72 in Afghanistan.[2]
Two production systems carry the region. Pakistan's dairy herd produced 72.34 million tonnes of milk in July to March 2024-25, with livestock contributing 63.6 per cent of agricultural value added and 14.97 per cent of GDP.[14][15] Bangladesh produced 5.02 million tonnes of fish in 2023-24, of which 2.978 million tonnes, 59.3 per cent, came from inland aquaculture.[10][11] FAO's 2026 assessment places Bangladesh among the world's top five aquaculture producers and second globally in inland capture,[12][13] and fish supplies roughly 60 per cent of Bangladeshi animal protein.[11] Across the seven members, farmed output reached about 3.19 million tonnes and capture about 3.09 million tonnes in 2023.[8][9]
The gaps are in cereals, pulses and feed. Bangladeshi wheat production of 1.05 million tonnes covers under 10 per cent of a 7.7 million tonne consumption.[29] Pakistan closed the 2026 wheat year with a 29.61 million tonne harvest and an estimated 3.5 million tonne shortfall, against a USDA projection of 27.5 million tonnes and 1.7 million tonnes of imports.[16][18] Bhutan's cereal self-sufficiency was 34.2 per cent in 2024,[51] the Maldives imports roughly 90 per cent of its food from 27 square kilometres of cultivable land,[53] and Nepal's food import bill reached Rs434.26 billion in FY2025-26.[49]
Durable domestic capacity: 56
Quantitative score 60, adjusted down 4 points
The indicators put the pillar at 60; we deduct 4 for Afghanistan's crisis-degraded capacity, giving 56.
The capacity is real where it exists. Pakistani milk output per head exceeds India's, and Bangladeshi pond aquaculture has become a globally significant industry while the country remained self-sufficient in rice at a forecast 37.4 million tonnes milled.[30] Sri Lanka reached self-sufficiency in poultry meat and eggs by 2024 after importing table eggs in 2022 for the first time since 1962,[48] and its fisheries supply over half of national animal protein.[45]
The limits are equally clear. Sri Lankan milk covers 43 per cent of daily requirement.[46] Undernourishment ranges from 5.3 per cent in Nepal through 10.4 per cent in Bangladesh and 16.5 per cent in Pakistan to 28.1 per cent in Afghanistan.[7] Afghanistan warrants the negative adjustment on its own terms: 17.4 million people, 36 per cent of the population, are projected into IPC Phase 3 or above with 4.7 million in Phase 4,[56] after four consecutive drought years and the failure of the 2025 rainfed wheat crop, with 885,400 people forcibly returned from Iran and Pakistan in the first half of 2026.[57] Active crisis degrading nominal capacity is a legitimate basis for the adjustment, and Afghanistan carries 8.4 per cent of regional population weight.
Resource headroom: 34
Quantitative score 33, adjusted up 1 point
The indicators put the pillar at 33; we add 1 for Bangladesh's demonstrated aquaculture capability, giving 34.
There is very little room to grow, and the trend is worse than a single year's stock figures show. Pakistan withdrew 109.99 per cent of its internal renewable freshwater in 2022 against an internal endowment of 225.69 cubic metres per person, and Sri Lanka withdrew 90.79 per cent.[4][5] The Indus basin supplies 96 per cent of Pakistan's available freshwater, roughly 80 per cent of it originating abroad, behind storage capacity of about 30 days.[62] Regional cropland grew just 0.1 per cent and arable land fell 1.6 per cent in absolute terms between 2000 and 2024 while population grew 46.9 per cent, so cropland per person fell 31.8 per cent and arable land per person fell 33.0 per cent;[80][82] six of the seven members lost arable land per capita, Sri Lanka the sole exception at plus 34.1 per cent on genuine absolute expansion of both cropland and irrigated area.[80] Irrigation infrastructure tells the same story rather than offsetting it: Pakistan has 65.1 per cent of its arable land equipped for irrigation, up from 58.3 per cent in 2000 on a real 10.5 per cent absolute expansion, while Bangladesh's equipped area, at 110.6 per cent of arable land and 99.0 per cent of the broader cropland base, is effectively saturated.[80] Bangladesh has the opposite land problem to Pakistan's water problem: abundant surface water at a 5.72 per cent withdrawal rate and 0.05 hectares of arable land per person,[4][6] an endowment that is shrinking. The salinity-affected area grew 26.7 per cent between 1973 and 2009, from 833,450 to 1,056,190 hectares,[64] with a further 8 to 12 per cent salinity increase recorded in parts of Satkhira and Khulna between 2024 and 2025.[65]
The marine base is fished out at both ends of the region. The World Bank assesses the 14,600 square kilometres of commercially significant grounds on Bangladesh's shelf, 12 per cent of the total, as overexploited, with silver pomfret biomass down 58 per cent since 2000 and Indian salmon down 63 per cent.[63] Pakistani marine capture has been flat at about 0.51 million tonnes since 2020,[9] with 60 to 90 per cent of stocks reported fully exploited or overexploited.[71] Sri Lankan capture fell from 382,266 tonnes in 2020 to 338,161 in 2023.[9]
Capital is not the binding constraint. FAOSTAT's agriculture orientation index for credit carries a reading for all seven members, Sri Lanka's the only one above parity at 1.17, Bangladesh's at 0.41 and Pakistan's at 0.22,[81] a materially better-capitalised picture than regions where the series is largely absent. Pakistan's own flagship expansion scheme shows what the ceiling looks like in practice: two years after the Green Pakistan Initiative's February 2025 Cholistan desert launch, backed by Rs211.34 billion of canal investment on the Sulemanki Barrage, investors including Unity Food Group and Fatima Fertilizer, 50,000 acres each, are withdrawing or have not started, one farm's groundwater salinity more than doubled after a failed wheat sowing, and surviving operations grow mostly Rhodes grass and alfalfa for export rather than food, against Indus flows at the Kotri gauge already down 65 per cent since the 1970s.[83][84]
The offsetting feature is a low-feed protein mix: Bangladesh directs 34.22 per cent of cereals to animals, but Pakistan only 17.43 per cent, Sri Lanka 7.82 and Afghanistan 3.45,[3] so each unit of protein consumes less primary resource than the feedlot systems of wealthier entities. Bangladesh extracts 2.978 million tonnes of farmed protein from the region's second-smallest arable base,[11] a demonstrated operating capability that earns a small uplift on its own. Nepal and Bhutan's water, at 6,670 and 99,883 cubic metres per person,[5] does not earn the same credit: Nepalese arable land fell 23.4 per cent between 2000 and 2024 as an estimated 2,000 to 2,500 agricultural workers a day left for foreign employment and roughly 100,000 hectares of government-owned arable land now sits abandoned,[85] and Bhutanese arable land fell 35.2 per cent over the same period while its credit orientation index has fallen by more than half from a 2017 peak.[80][81] Water that is unexploited because the land and labour to use it are disappearing is not room to grow, which is why the uplift is capped at 1.
Import exposure: 47
Quantitative score 52, adjusted down 5 points
The indicators put the pillar at 52; we deduct 5 for supplier concentration the volume ratios miss, giving 47.
Staple protein is mostly domestic, so exposure concentrates in categories substitutable in theory and unsubstitutable in practice. Pakistan imported roughly 1.32 million tonnes of pulses in FY2025, near 80 per cent of consumption, for close to USD 1 billion as domestic production fell about 40 per cent,[66] plus around 2 million tonnes of soybeans for crush after the GM import ban was lifted in November 2024.[67] Bangladesh imports 6.7 to 7.4 million tonnes of wheat,[29] 1.5 million tonnes of maize[30] and 2.4 million tonnes of soybeans[31] a year, and produces 175,000 tonnes of lentils against a requirement of 600,000 to 700,000.[34] Sri Lanka buys roughly 100,000 tonnes of milk powder annually for over USD 400 million,[46] licences maize imports against a 600,000 tonne feed requirement,[47] and spent USD 90.3 million on rice imports in the first five months of 2026.[42]
The three smallest members are import-fed by construction. Food accounts for 21.24 per cent of Maldivian merchandise imports, with staples moving under quota.[53] India supplied 90 to 95 per cent of Bhutanese agricultural imports by volume and value in 2024.[51] Nepal imported Rs39.53 billion of rice and paddy, Rs25.77 billion of live animals, poultry and dairy, and Rs34.03 billion of oil cakes and maize for feed in FY2025-26.[49]
The downward adjustment reflects concentration that volume ratios miss. Close to 100 per cent of Nepalese and Bhutanese rice imports come from India,[70] which is also the marginal supplier for Bangladesh, Sri Lanka and the Maldives. That supplier imposed a wheat export ban in May 2022 and a non-basmati white rice ban from July 2023 to September 2024, and it is unavailable at any price to Pakistan. Afghanistan compounds the problem from the other direction, landlocked and reliant on overland Kazakh, Uzbek and Turkmen wheat as Kazakhstan's exportable surplus falls to 7.5 million tonnes in 2026/27.[57]
Upstream dependence: 26
Quantitative score 23, adjusted up 3 points
The indicators put the pillar at 23; we add 3 for Pakistan's funded shift of urea capacity onto domestic gas, giving 26. This is the region's defining weakness and the pillar the 2026 fertiliser shock has struck hardest.
Bangladesh needs 2.6 million tonnes of urea a year and produces 800,000 to 1,000,000 tonnes, under a third, down from 1.9 million tonnes and 80 per cent of demand.[25] Four of five BCIC plants plus KAFCO shut from 4 March 2026 on gas shortage, leaving one plant at 2,800 tonnes a day, after which the country failed twice to buy replacement volumes: a first tender drew zero bids and a second drew 50,000 tonnes against 200,000 sought.[25] Its DAP demand of 1.5 to 1.6 million tonnes is met by 100,000 tonnes of domestic output and a 600,000 tonne Saudi contract, about a third of imports, whose March 2026 consignment was cancelled.[26][27] All potash is imported, with 859,000 tonnes planned for 2026.[26]
Pakistan is nominally urea self-sufficient, on gas it does not have. It imported 6.6 million tonnes of LNG in 2025, almost all Qatari, and 35 cargoes for 2026 delivery were cancelled before the crisis.[24] Its DAP demand of 1.3 to 2.3 million tonnes is served by a single 650,000 tonne plant, with 271,000 tonnes imported against 381,000 produced in the Rabi season from October 2025 to March 2026.[19][20]
Price transmission has been rapid. Roughly a third of global seaborne fertiliser trade, about 16 million tonnes, transits Hormuz.[38] Urea passed USD 850 per tonne in April 2026, 80 per cent above February and the highest since April 2022; potash is 17 per cent higher year on year, and the World Bank projects its fertiliser index rising more than 30 per cent across 2026.[35] FAO's point is that the crop calendar cannot be deferred, so the yield consequences land in the second half of 2026 and into 2027.[36] Bangladeshi urea moved from USD 484 to 690 per tonne between February and early April, with LNG freight rates up 650 per cent.[28]
Sri Lanka has already run this experiment to conclusion. The April 2021 ban on chemical fertiliser and pesticide imports cut rice yields by around 30 per cent, forced USD 450 million of rice imports into a country that had been self-sufficient, and raised domestic rice prices roughly 50 per cent.[39][40] The subsidy that replaced it cost LKR 37 billion in 2025.[41]
The upward adjustment recognises one funded structural change. Pakistan allocated 222 mmcfd of indigenous gas to three fertiliser plants in September 2025, confirmed by federal cabinet in December, covering the roughly 32 per cent of urea capacity, over two million tonnes a year, that previously lacked assured feedstock.[21] It is capped at 3 because no phosphate rock or potash exists anywhere in the region, and because the Indus Waters Treaty abeyance leaves the water itself under another state's control.[61]
Access and affordability: 38
Quantitative score 40, adjusted down 2 points
The indicators put the pillar at 40; we deduct 2 for Afghanistan's absence from the affordability series, giving 38.
Restricting to the six members with World Bank/FAO Cost and Affordability of a Healthy Diet data, 91.6 per cent of regional weight, the population unable to afford a healthy diet in 2024 ranges from 4.1 per cent in Bhutan and 4.7 per cent in the Maldives to 20.0 per cent in Nepal, 36.3 in Sri Lanka, 40.4 in Bangladesh and 64.4 in Pakistan, a population-weighted share of roughly 51.5 per cent.[60] A healthy diet costs between 3.94 and 6.17 PPP USD per person per day across the four members with a published 2025 figure, cheapest in Pakistan and dearest in Bhutan.[59] Undernourishment averages roughly 14 per cent, weighted across the five members with data and covering 99.7 per cent of regional weight, from 5.3 per cent in Nepal to 28.1 in Afghanistan.[7] Stunting is the sharpest outcome check and the one affordability-adjacent series that does cover all seven members: a population-weighted average of about 32 per cent sits inside the World Health Organization's "very high" band, from 10.5 per cent in Sri Lanka to 37.6 in Pakistan and 44.6 in Afghanistan.[73] Prices have swung rather than merely risen: Sri Lankan food inflation moved from 94.9 per cent in September 2022 to minus 5.2 exactly twelve months later,[76] and Pakistani food inflation from 48.7 per cent in May 2023 to minus 5.1 in March 2025,[74] with Bangladesh[75] and Nepal[77] tracking narrower swings. Physical access is constrained by conflict and displacement in Afghanistan,[57] dispersed-atoll sea-only supply in the Maldives,[53] cyclone and salinity exposure on the Bangladeshi[65] and Sri Lankan[41] coasts, and Pakistan's roughly 30-day water storage buffer.[62]
The downward adjustment reflects Afghanistan's absence from the affordability series, the heaviest of this pillar's indicators. Its own undernourishment of 28.1 per cent, its 44.6 per cent stunting rate and 17.4 million people in IPC Phase 3 or above[56][73] all place its unmeasured unaffordability far above the six-member figure of 51 per cent, so the headline indicator understates the true position. The adjustment is held to 2 points rather than more because three of the pillar's other four indicators, undernourishment, stunting and physical access, already carry Afghanistan, and because correcting the affordability share to roughly 54 per cent leaves it within the same 45 to 65 per cent band. Afghanistan carries 8.4 per cent of population weight (-2).
Shock endurance: 39
Quantitative score 41, adjusted down 2 points
The indicators put the pillar at 41; we deduct 2 for the dismantling of Pakistan's strategic reserve institution, giving 39. This is the most divergent pillar in the region.
Bangladesh holds the strongest position: USD 32.93 billion in BPM6 reserves at June 2026, up from USD 26.74 billion a year earlier, giving about five months of cover,[32] alongside 1.52 million tonnes of public food stocks and a 2.29 million tonne rice procurement plan,[33] and 1.68 million tonnes of fertiliser stock at 6 March 2026 assessed as sufficient to May or June.[28] Nepal holds USD 23.55 billion and 21.8 months of merchandise import cover, built on remittance inflows,[50] purchasing power it cannot convert into domestic output.
Pakistan, at 48 per cent of population weight, holds USD 22.67 billion in total liquid reserves and 2.54 months of import cover, week ended 17 July 2026, under an IMF Extended Fund Facility whose third review, releasing a further USD 1.32 billion, completed on 8 May 2026.[22][23] It is dismantling PASSCO, holder of the 1.7 million tonne federal wheat reserve, in the same year that Punjab procured 480,000 tonnes against a three million tonne target, 5.1 million tonnes went unaccounted for and the cabinet decided to import a million tonnes at roughly USD 1 billion.[16][17] Sri Lankan reserves fell 6.2 per cent in June 2026 to USD 6.45 billion,[43] with over 92 per cent of external debt restructured but sustainability risks still assessed as high,[44] behind a strategic paddy reserve of 49,612 tonnes.[41] Maldivian usable reserves fell from USD 113.6 million to USD 73 million between end-April and end-May 2026, barely a month of imports, with gross cover down to 1.4 months in April after Sukuk repayment.[54][55] Afghanistan has no capacity of this kind: WFP reaches 2.1 million people a month with about 3,700 tonnes, requires USD 350 million to October 2026, and no food assistance is planned across the current projection period.[57][58]
Crisis purchasing power, foreign exchange reserves in months of imports, fiscal and market access, and social protection able to scale, widens rather than narrows this divergence. Pakistan's Ehsaas Emergency Cash programme reached 13.2 million people with PKR 160.4 billion disbursed by July 2020, a real but now dated precedent set against today's 2.54 months of import cover.[78] Bangladesh's WFP-supported safety net reached only around 4 million people in 2024,[79] against a population of about 178 million,[72] five months of reserves and a rice surplus. Afghanistan has no scalable social protection system of its own at all.
The downward adjustment reflects the deliberate dismantling of the region's largest strategic reserve institution at the worst possible moment. Reserve stocks are the mechanism this pillar measures, and Pakistan is winding up PASSCO, its only federal holder of one, in the year its provincial procurement collapsed and 5.1 million tonnes went unaccounted for,[16][17] with 2.54 months of import cover and an IMF programme constraining the fiscal response.[22][23] That is a loss of absorptive capacity the reserve and stock indicators do not register, and it falls on 48 per cent of regional consumption. Partly offsetting it, Sri Lanka has rebuilt milled rice output to a forecast 3.45 million tonnes from a self-inflicted input collapse[41] and Nepal holds 21.8 months of cover.[50] Affordability is not scored here: the share of the population priced out of a healthy diet is steady-state distributional performance and is scored once, in access and affordability above. What belongs to this pillar is the transmission channel, and the 2026 shock has already added 30 per cent to the grains and oilseeds freight index since 28 February.[37]
Trajectory
Three changes carry the most weight. The first is domestic nitrogen: Pakistan's Mari gas allocation is the right template and Bangladesh needs its equivalent, because a urea system running at under a third of demand behind one functioning plant is the clearest single failure point. The second is phosphate and potash routing away from Hormuz, where the Egyptian, Moroccan, Chinese and Emirati alternatives Bangladesh began negotiating in April 2026[27] are a beginning rather than a solution. The third is aquaculture feed: Bangladesh's fish output is the region's best protein asset and it rests on imported soybean meal, imported fishmeal from the Maldives and Oman, and fish oil from Chile and Norway.[68][69]
Working against all three is water. Pakistan's overdraft is structural, its storage is 30 days, and the treaty governing four-fifths of its supply is suspended. Bhutan targets self-sufficiency by 2029[52] and the Maldives is building climate-resilient agriculture across 26 northern islands,[53] both worth doing and neither capable of changing this picture. The direction of travel will be set by whether the fertiliser shock is absorbed in one harvest or two.
Country notes
Pakistan. The member whose weaknesses set the score. Large domestic production, 72.34 million tonnes of milk, against 110 per cent freshwater withdrawal, 2.54 months of import cover, a collapsed 2026 procurement round and the dismantling of its only federal grain reserve institution.
Bangladesh. The positive outlier. Rice self-sufficient, a top-five aquaculture producer, five months of reserve cover and 1.52 million tonnes of public stocks, all resting on 0.05 hectares of arable land per person, imported wheat, imported soy and a urea industry that stopped in March 2026.
Sri Lanka. The cautionary tale and the recovery case. Rice output remains below pre-2021 levels four years after the fertiliser ban and milk covers 43 per cent of requirement, but poultry and eggs are self-sufficient and the IMF programme is on track.
Nepal. Twenty-two months of import cover and a Rs434 billion food import bill. Remittances have bought food security and hollowed out the farm labour force that might have produced it, arable land fell 23.4 per cent between 2000 and 2024, and near-total reliance on India for rice makes one supplier the single point of failure.
Bhutan. Cereal self-sufficiency of 34.2 per cent and 90 to 95 per cent of agricultural imports from India, offset by full egg self-sufficiency and 92 per cent in dairy. It holds the region's largest water endowment per person, but arable land has fallen 35.2 per cent since 2000 and its agricultural credit orientation has been falling for seven straight years.
Maldives. Roughly 90 per cent of food imported onto 27 square kilometres of land, and the region's highest protein supply at 97.41 grams per person per day thanks to 160,683 tonnes of capture fisheries. The constraint is foreign exchange, at about one month of usable reserves.
Afghanistan. At or near the floor on every pillar: undernourishment of 28.1 per cent, 17.4 million people in IPC Phase 3 or above, four drought years, 885,400 forced returns in six months, and no planned food assistance.