Definition
Protein Sovereignty: a population's ability to sustain affordable access to enough protein to meet its nutritional needs, even through disruption to production, trade and critical inputs.
The index distinguishes protein production from protein sovereignty. A country may produce heavily and remain fragile, or import heavily and remain resilient.
Unit of analysis
40 entities provide complete world coverage: 24 individual countries, 2 blocs assessed as single units (the European Union and the Gulf Cooperation Council), and 14 regions covering all remaining countries. Regional scores are population-weighted judgements across member states; every sovereign state is named and characterised within its region's assessment. Dependencies and territories are coloured with the geographic region they sit in.
The six pillars
Each pillar is scored 0–100 (higher = more sovereign). The pillars run in causal order: what a country can produce, what that production could become, what it must bring in, what its own production secretly depends on, whether its people can actually afford the result, and what happens when the system is hit.
- Durable domestic capacity: How much of the population's protein needs can the country produce at home, and can that production be sustained without exhausting the land, water, fisheries or other biological resources on which it depends?
- Resource headroom: How much land, water, feed and other primary resource does each unit of protein consume, and does the resource base have room left to grow production the country could realistically mobilise?
- Import exposure: How much protein must be imported, and how concentrated are those flows among particular suppliers, ports and maritime chokepoints?
- Upstream dependence: How much of domestic production relies on imported feed, fertiliser, energy, genetics, medicines, equipment or other critical inputs? How concentrated and replaceable are those dependencies?
- Access and affordability: Does the combination of domestic production and imports provide people across income groups and regions with reliable access to enough nutritionally adequate protein, at prices they can afford and that hold steady?
- Shock endurance: How long can the system maintain affordable supply when domestic production, critical inputs or trade are disrupted? How readily can it draw on reserves, reroute supply, substitute inputs, adapt and recover?
Pillar boundaries
Adjacent pillars share subject matter; these seams assign each concern a home. A single piece of evidence may inform more than one pillar where it answers each pillar's distinct question: an aquifer being drawn down both undermines the durability of current output and shrinks the room to grow, and each pillar scores what the evidence says about its own question. What the seams prevent is the same concern being scored twice for the same reason.
- Capacity vs headroom: the sustainability of current production (aquifer drawdown, soil degradation, stock collapse) is priced in capacity; resource efficiency and room to expand belong to headroom.
- Headroom realisability: expansion potential is credited only where the entity has a plausible path to mobilising it (capital, infrastructure, land governance, security). Pure agronomic potential without that path is disclosed in the assessment, not scored.
- Imports vs upstream: import exposure counts protein arriving as finished food; inputs to domestic production (feed, fertiliser, genetics) count under upstream dependence, so the same flow is never scored in both.
- Upstream measures dependence, not productivity: a low-input production system scores well on upstream dependence by construction, because dependence that does not exist cannot be cut off. Low yields and fragile output belong to capacity; hardship in consumption belongs to access and affordability. A poor entity can therefore outscore a rich one on this pillar, and that is the pillar working as defined. Where the low-input position reflects inability rather than choice, with no domestic input manufacture and no substitution route if inputs were ever needed at scale, the score takes a bounded deduction (four to six points in this edition), logged like any other adjustment, so that poverty is not read as independence.
- Access vs endurance: steady-state performance (distribution across income groups and chronic price volatility) belongs to access and affordability; behaviour under disruption (reserves, rerouting, substitution, fiscal and foreign-exchange firepower) belongs to shock endurance.
Scoring: quantitative core + bounded judgement
Each pillar score starts from named quantitative indicators, then admits a bounded qualitative adjustment of at most ±15 points, each logged with a written justification in the public adjustment log. Unadjusted quant scores and adjustments are both published.
Core indicators by pillar
| Pillar | Primary indicators | Principal sources |
|---|---|---|
| Durable domestic capacity | Protein self-sufficiency ratio (production ÷ domestic supply, total and animal protein); per-capita protein supply vs requirement; sustainability of the production base (water stress, soil, fish-stock status) | FAO Food Balance Sheets; FAOSTAT production; AQUASTAT; FAO SOFIA |
| Resource headroom | Feed-conversion profile of the protein mix; share of protein from feed-intensive systems; wild-fishmeal reliance of aquaculture; arable land and freshwater headroom, discounted for realisability; share of crop calories fed to livestock | FAOSTAT; FAO SOFIA; peer-reviewed feed-conversion literature |
| Import exposure | Protein import dependency ratio; supplier concentration (HHI across partner countries for soy, fishmeal, meat, dairy, pulses); share of imports transiting major maritime chokepoints (Hormuz, Malacca, Panama, Suez, Bab el-Mandeb, Bosphorus) | UN Comtrade; FAOSTAT trade matrix; Chatham House resourcetrade.earth |
| Upstream dependence | Import dependence for feed protein (soymeal/fishmeal), nitrogen/phosphate/potash fertiliser, agricultural energy, breeding genetics and veterinary inputs | FAOSTAT; UN Comtrade; IFA fertiliser statistics |
| Access and affordability | Cost of a nutritionally adequate diet relative to bottom-quintile income; share of the population unable to afford a healthy diet; prevalence of undernourishment and child stunting as outcome checks; domestic food price volatility; physical access constraints (infrastructure, conflict-affected regions) | FAO SOFI; World Bank CoAHD and ICP; FAO food price monitoring; WFP |
| Shock endurance | Stock-to-use ratios and strategic reserves; substitution capacity (dietary breadth, alternative feed routes); fiscal and foreign-exchange capacity to outbid on world markets; social protection systems that can scale in a crisis; recent shock performance (COVID, Ukraine grain shock, avian influenza); policy preparedness | USDA PSD; national reserve disclosures; World Bank; IMF reserve data; documented shock history |
Indicator → score mapping
Each pillar score is set from the named, cited indicators listed in its assessment, against the score-band definitions published in the research rubric. Two components are computed directly against published anchor tables: the access and affordability pillar, anchored on the share of the population unable to afford a healthy diet with the full mapping in the rubric, and the crisis purchasing power component of shock endurance, with its scale in the annex. The remaining pillar scores are analyst judgements disciplined by the published band definitions, cross-entity consistency review and the sourcing standard below; they are evidence-led, not formula-generated, and the index does not claim otherwise. Scores are comparable across editions because the band definitions and anchor tables are fixed and versioned. Data year: most recent available, 2022–2025; the data year for every figure is cited.
Qualitative adjustment rules
Adjustments capture what data misses, within ±15 per pillar. Examples of admissible grounds: offshore farmland holdings with secured logistics (GCC); demonstrated wartime adaptation (Ukraine); active conflict degrading nominal capacity; credible near-term policy shifts with funding attached. Inadmissible: announced targets without funding, national pride, vendor claims. Every adjustment appears in the public log: entity, pillar, direction, size, justification, sources.
Overall score and tiers
Overall = weighted mean of the six pillars. The pillars make different kinds of claims: three measure structure, two measure performance, one estimates potential, and the weights reflect how central and how certain each claim is. They were fixed before any entity was scored under this framework.
| Pillar | Weight | Rationale |
|---|---|---|
| Durable domestic capacity | 25% | The anchor of the definition: produce at home, sustainably. |
| Resource headroom | 10% | A projection about expansion rather than a measurement of the present; real, but the least certain pillar. |
| Import exposure | 20% | The direct measure of dependence on suppliers, ports and chokepoints. |
| Upstream dependence | 15% | The same dependence one step removed: it hollows out domestic capacity rather than replacing it. |
| Access and affordability | 15% | The outcome check: sovereignty that leaves people unable to afford protein is not sovereignty; weighted so the index measures sovereignty rather than becoming a general food-security index. |
| Shock endurance | 15% | "Even through disruption" is in the definition verbatim. |
In short: structure carries 60% (capacity, imports, upstream), performance 30% (access, endurance), potential 10% (headroom). As a robustness check, calibration recomputes every overall score under equal weighting and any entity that changes tier is disclosed in the methodology annex.
Tier thresholds are inclusive: a score landing exactly on a boundary takes the higher tier.
| Tier | Score | Meaning |
|---|---|---|
| Sovereign | 80–100 | Can meet protein needs domestically or through robustly diversified supply, sustainably and affordably, through extended disruption |
| Resilient | 65–80 | Strong capacity or well-diversified supply with manageable dependencies |
| Exposed | 50–65 | Material dependencies on concentrated suppliers, chokepoints or imported inputs |
| Dependent | 40–50 | Structural reliance on imports or imported inputs with limited substitution capacity |
| Critically exposed | 0–40 | Severe concentration of supply through few suppliers/chokepoints, little endurance or affordability, or an eroding resource base |
Overall scores are computed to one decimal place, rounding half up, before banding. Each band includes its lower bound and excludes its upper: an overall of exactly 65.0 is Resilient, and an overall of exactly 40.0 is Dependent.
Sourcing standard
Every quantitative claim in an assessment carries a numbered citation to a primary or authoritative secondary source (FAO, UN Comtrade, USDA, World Bank, national statistics, peer-reviewed literature, or named quality journalism for events), with its data year stated.
Methodology annex
Robustness to weighting. As a check on the published weights, every overall score is recomputed under equal weighting of the six pillars. In PSI 2026 two entities change tier under equal weighting, both downward across a band boundary they sit close to. Ethiopia (41.1, Dependent) falls to 39.0, into Critically exposed: its strongest pillars (import exposure 55, capacity 50) carry more weight than its weakest (access 19), so the published weighting reads its structure as slightly more sovereign than an unweighted view would. Ukraine (65.0, Resilient) falls to 63.2, into Exposed: it enters Resilient exactly on the boundary, and equal weighting gives more voice to its war-damaged access (50) and upstream (48) pillars than the published weights do. Every other entity keeps its tier. We read this as the weights mattering at the margins, not driving the results; both divergences sit within a band's width of the boundary, and the direction of each is explained by the entity's own pillar spread rather than by any systematic lean in the weighting.
Distribution and regional averaging. The 2026 edition places 1 entity in Sovereign, 8 in Resilient, 14 in Exposed, 13 in Dependent and 4 in Critically exposed. The extremes are thinner than a country-level reading of the world would suggest because regional entities are population-weighted aggregates: countries in famine or blockade (Sudan, Yemen, North Korea) sit inside regional averages that better-functioning neighbours pull upward. The per-country picture appears in each regional assessment's member notes.
Anchor integrity. The access pillar anchors on the FAO/World Bank cost and affordability of a healthy diet series. Where that series is suppressed for a country, fails internal consistency between its prevalence and headcount releases, or rests on income distributions that cannot describe the population being scored, the anchor is discarded and the pillar is scored on measured current evidence with the provenance of every input stated. No figure is interpolated. The European Union's affordability reading additionally draws on Eurostat measures alongside the series used elsewhere, and is disclosed as such in its assessment.
Resource headroom realisability. Expansion potential is credited only on collected evidence: irrigation-equipped share of arable land and its trend, agricultural credit orientation (with missing series recorded as findings), and cropland change in absolute and per-capita terms together with the documented outcomes of flagship expansion programmes. Occupied, mined or otherwise structurally unreachable resource is excluded from the quantitative score rather than discounted by adjustment.
Crisis purchasing power within shock endurance. The endurance pillar prices the capacity to keep buying through a crisis at 20 per cent of its quantitative core, on a single scale applied to all entities: foreign-exchange reserves in months of import cover (piecewise from 5 points at zero cover to 97 at two years or more), a floor of 60 for reserve-currency issuers, a 15-point credit for sovereign wealth of macro-decisive scale, and modifiers for credit-market access and crisis-scalable social protection. Purchasing power is discounted where a documented chokepoint closure leaves a material share of an entity's food imports with no substitute route at scale, because purchasing power cannot reopen a strait; in this edition no entity meets that test, in each case because substitute routes are documented.
The judgement layer. Every pillar score is a quantitative core plus a bounded qualitative adjustment of at most 15 points in either direction, each logged with a written justification. The 2026 edition carries 240 adjustments: 110 negative, 93 positive, 37 zero. The bound binds nowhere.
Precision. Scores are reported to one decimal to make the ranking reproducible, not to claim decimal accuracy. Several adjacent entities are separated by less than one point; the table ranks scores, not certainties, and the essays state the margins where they are material.
Editions
Scores are published in annual editions; PSI 2026 is the first. When the methodology changes between editions, the change is documented so that score movements are attributable to the world, not the model.