Skip to main content

​
Start here

The five-factor scorecard asks five blunt questions about a country, and answers each one with a number from 1 to 5:
  1. Food — can it feed itself?
  2. Energy — can it supply its own energy?
  3. Demographics — does it have the people, skills, and workforce to sustain itself?
  4. Technology — can it develop and use technology?
  5. Defense — can it defend itself?
A 1 means a severe structural deficit. A 5 means the country is largely self-sufficient on that factor. Nothing else about the number is complicated. The point is to answer, in one screen, the question analysts usually spend a week assembling from a dozen datasets: if this country were cut off, what would break first?
The scorecard describes structure, not events. It changes on the scale of years, because the underlying data — harvests, energy balances, census age structures, defense budgets — is published annually. For fast-moving risk, use the Composite Instability Index instead.

​
How to read a score

Each score is absolute, not a ranking. A 4 for Japan and a 4 for Brazil mean the same thing about self-sufficiency; they are not “4th place.” A country can score 5 on every factor, or 1 on every factor, and both are legitimate outcomes. Nothing is graded on a curve. Alongside the 1-5 score, every factor also reports a sub-score from 0 to 100. Use the 1-5 score to compare countries at a glance; use the sub-score when you need to see movement inside a band — the difference between a country sitting at 61 (a shaky 4) and one sitting at 79 (nearly a 5).

​
A worked reading

Suppose a country returns:
  • Food 4 (sub-score 68.8)
  • Energy 2 (sub-score 31.0)
  • Demographics 3
  • Technology 4
  • Defense 2
The plain reading: this country grows more food than it eats and holds decent reserves, but it buys most of its energy abroad and depends on foreign suppliers for military equipment. A blockade or a sanctions regime would hurt it through fuel and weapons long before it hurt anyone’s dinner plate.

​
What the scorecard is not

  • Not a quality-of-life or “good country” index. A wealthy, safe, deeply interdependent country can score low. Self-sufficiency and desirability are different things.
  • Not a forecast. It says what capacity exists today, not what will happen.
  • Not the Composite Instability Index (CII), which measures near-term instability risk, and not the Country Resilience Index (CRI), which measures recovery capacity. The three are independent and are not substitutes for one another.
  • Not editorial. Every number traces back to a named public source observation with a year attached, which the API returns alongside the score.

​
How a score gets built

Every factor is built the same way, in four steps. Step 1 — collect the observations. For each factor, the scorer pulls a small set of published indicators. Food, for example, uses calorie production, calorie consumption, ending stocks, water stress, and import concentration. Step 2 — put every indicator on the same 0-100 ruler. Raw indicators arrive in incompatible units — tonnes, percentages, dollars, people. Each one is converted onto a common 0-100 scale using two fixed reference points, called goalposts: a value that scores 0 and a value that scores 100. Anything at or beyond a goalpost is clamped to that end. For example, the food-balance goalposts are 0.50 -> 0 and 1.25 -> 100. A country producing half the calories it consumes scores 0; one producing 25% more than it consumes scores 100; one producing exactly as much as it consumes lands at 67. The goalposts are frozen as part of the published methodology, so last year’s score and this year’s score are measured against the same ruler. Step 3 — take a weighted average. Indicators are not equally important. Within Food, the production-versus-consumption balance carries 55% of the weight, while import diversity carries 5%. The weighted average of the component scores becomes the factor’s 0-100 sub-score. Step 4 — convert to a 1-5 score. The sub-score is mapped to a band using fixed cutoffs. The lower boundary is inclusive, so exactly 40.0 is a 3.

​
Worked example: one country’s food score

Take a country with these four published observations. All four indicators are present, so coverage is 1.00 and the contributions simply add up:
Read as prose: the country grows about 5% more calories than it eats, holds roughly two months of buffer stock, has moderate water stress, and spreads its food imports across enough partners that no single supplier can squeeze it. Strong capability, with room to improve on buffers.

​
Why a factor sometimes has no score at all

Sometimes a factor comes back empty instead of low. That is deliberate, and the distinction matters: a missing score is not a bad score. The scorer never invents a placeholder for a missing indicator. It does not substitute a neutral 50, and it does not treat “we have no data” as zero capability — doing either would quietly turn a data gap into a false finding about a real country. Instead, each factor has a coverage floor: a minimum share of its indicator weight that must actually be present, plus a short list of indicators it cannot do without. Food, for instance, needs 70% of its weight available and must have the production-versus-consumption balance. If a country clears the bar, the factor is scored using only the indicators that are present, with the weights rescaled across them. If it does not clear the bar, the factor reports no score and says why.
Never read a missing score as a zero. Over the API, unscored factors come back with hasScore: false and numeric fields set to 0 — those zeros are protocol placeholders for “no data,” not measurements. Always check hasScore before reading score or subScore. See Reading the API response.
Every unscored factor names its reason from this closed set:

​
How current the data is

Each indicator carries a maximum age, because a 12-year-old energy balance is not evidence about today. The boundary year is inclusive. An observation past its limit is marked stale and drops out of the score rather than dragging it. If a source’s own content-age envelope expires, every indicator it feeds goes stale too, even when the stored values are still readable. The scorecard also refuses to publish a thin day. A fresh daily cohort replaces the previous one only when it covers at least 180 countries with a scoreable factor and 150 with usable population evidence, and clears these per-factor country counts: food 80, energy 120, demographics 150, technology 110, defense 30. These floors came from an audit of 196 countries against production sources, with headroom for normal coverage wobble. The pre-activation production refresh measured 116 scoreable technology countries, so that floor was corrected to 110 — a six-country outage margin — without loosening any country’s evidence, scoring, freshness, or null rules. A partial source outage therefore cannot overwrite a richer snapshot with a poorer one — the previous good cohort simply stays live.

​
The five factors in detail

Each factor below lists the indicators it uses, how much each one counts, the 0-100 goalposts, and how the score is aggregated when you ask for a bloc rather than a single country.

​
Food

The question: can the country cover the calories it consumes from its own production and stored reserves, without being at the mercy of one supplier? Food combines a physical calorie balance with buffer stocks, water stress, and import concentration. Commodity quantities published in thousand metric tonnes are converted to trillion kcal as thousand metric tonnes * kcal/kg / 1,000,000 using a frozen commodity conversion table before anything is aggregated. Coverage floor: 0.70. Calorie production / consumption is required.
  • Score 1: production is near or below half of use and buffers are weak.
  • Score 3: domestic output covers much, but not all, use — or buffers are mixed.
  • Score 5: output materially exceeds use and stock buffers are strong.
For blocs: production and consumption are summed across members before their ratio is scored, and ending stocks and total use are summed the same way — the bloc is treated as one physical system. Water and import diversity are population-weighted across the members that have evidence.

​
Energy

The question: does the country produce the primary energy it burns, and is the system that delivers it sound? Consumption comes from OWID primary_energy_consumption. Production is derived from the audited net-energy-imports observation already carried by the resilience static source, Eurostat nrg_ind_id for covered European countries and World Bank EG.IMP.CONS.ZS elsewhere:
Both providers are audited for raw redistribution, so the netEnergyImportsPercent observation carries the provenance of whichever one supplied it. Coverage floor: 0.60. Primary production / consumption is required.
  • Score 1: production covers little of use.
  • Score 3: the balance is mixed and imports remain material.
  • Score 5: energy self-sufficient or a net producer, with strong supporting power-system evidence.
For blocs: production and consumption TWh are summed before scoring. Low-carbon share and grid efficiency are population-weighted across members with evidence.

​
Demographics

The question: does the country have enough working-age people, and are they educated and trained enough to run a modern economy? This factor deliberately mixes how many people are available to work with what they can do — a favorable age pyramid with no engineers is not capability, and neither is a deep university system attached to a collapsing workforce. Inputs come from demographics:capability:v1. Coverage floor: 0.60. At least one age-structure input and one education, research, or workforce input are required.
  • Score 1: severe dependency or contraction, with little capability evidence.
  • Score 3: mixed age structure and human-capability depth.
  • Score 5: favorable labor supply plus deep education, research, and industrial workforce capacity.
For blocs: the population-weighted mean of member sub-scores. It never averages the 1-5 scores — averaging bands would let a tiny member swing the result as hard as a large one.

​
Technology

The question: is the country connected, and does it generate its own technology rather than only consuming it? The existing technology-readiness score, rank, and components are unchanged; v1 adds the raw observations needed to show and reproduce this scorecard. Coverage floor: 0.65. At least one connectivity input and one innovation input are required.
  • Score 1: limited digital access and little measured innovation capacity.
  • Score 3: broad use or research capacity, but material gaps remain.
  • Score 5: high connectivity plus deep and sustained innovation capability.
For blocs: the population-weighted mean of member sub-scores.

​
Defense

The question: can the country sustain a military — and equip it without depending on someone else’s factories? Note the heaviest weight sits on the arms-transfer balance rather than on spending. A large budget spent entirely on imported equipment is a weaker structural position than a smaller budget backed by domestic production, and the weighting says so. World Bank defense observations come from military:industrial-base:v1. Coverage floor: 0.50. At least one posture input (spending or personnel) and the arms-transfer industrial-balance input are required.
  • Score 1: small posture with high external equipment dependence.
  • Score 3: material posture or industrial capability, with important gaps.
  • Score 5: large sustained posture and strong domestic and export industrial depth.
The industrial balance is only computed when exports and imports are both explicit, finite observations from the same year. A missing side is never coerced to zero; an explicit measured zero remains valid.
Supplier diversity is currently unavailable for every country. SIPRI raw transfer rows are not stored or returned, and until the source policy permits public use of the already derived supplier HHI, that component reports redistribution-blocked. Defense scores are therefore computed from the remaining 0.80 of weight.
For blocs: the population-weighted mean of member sub-scores, with supplier diversity still unavailable while the policy block applies.

​
Scoring a bloc instead of a country

You can score a group of countries as a single unit — the EU as one energy system, BRICS as one food system. Six presets ship as part of the versioned methodology: USMCA, EU27, BRICS, GCC, ASEAN, and NATO. Custom blocs take 2-30 unique uppercase ISO-2 codes from the public rankable universe. A request selects exactly one preset or one custom member list. Presets themselves may exceed 30 members. Two aggregation rules are used, and the difference is deliberate:
  • Food and energy are aggregated physically, then scored. Tonnes and terawatt-hours are summed across members first, because a bloc really does share one physical balance sheet.
  • Demographics, technology, and defense are population-weighted continuous scores. These are per-capita capabilities that cannot be summed, so member sub-scores are averaged by population.
No bloc formula ever averages the 1-5 scores. Each bloc factor reports its aggregation method, its included and excluded members, population coverage where applicable, and the evidence used. Preset membership was verified on 2026-08-29 against the official EU country list, BRICS member list, ASEAN member list, and NATO member-country history. The ASEAN preset includes Timor-Leste and the BRICS preset uses the official 11-member list. Membership changes require a methodology changelog entry.

​
For developers

​
Access

All three scorecard RPCs — GetFiveFactorScorecard, GetBlocScorecard, and ListFiveFactorScorecards — require a Pro subscription (tier 1), as do the MCP tools get_five_factor_scorecard and list_five_factor_scorecards. See Pro Intelligence Suite for request shapes.

​
Reading the API response

Each factor returns four fields that are easy to confuse: inputCoverage reports the share of indicator weight that was available, from 0 to 1, rounded to 4 decimals.
Protobuf JSON keeps numeric fields present as zero, so an unscored factor is indistinguishable from a genuine zero unless you branch on hasScore first. Those zeros are insufficient-data placeholders, not measured zero capability.

​
Do not re-derive the score from the sub-score

score is derived from the unrounded continuous value; subScore is that same value rounded to two decimals. Near a band boundary the two can legitimately disagree. A continuous value of 19.996 publishes as:
That is correct, not a bug. Always use the published score and band; never recompute a band from subScore.

​
Evidence records

Every indicator returns a tagged available/unavailable record. Available evidence carries the input ID, value, year, unit, source, and source key. A retained last-good upstream observation stays available and is marked quality=retained.

​
Version contract

  • Methodology: 1.0.0
  • Input registry: 1.0.0
  • Stored schema: 1
  • Canonical snapshot: scorecard:five-factor:v1
  • Atomic read model: scorecard:five-factor:v1:read-model
  • Seed health: seed-meta:scorecard:five-factor
Every result carries its methodology version and computation time. Changing a weight, goalpost, band cutoff, coverage floor, required group, aggregation rule, or input mapping requires a methodology version bump and a changelog entry. A change to the stored shape also requires a schema version bump. Health publishes the population count and the five per-factor counts separately.

​
Changelog

​
1.0.0 - 2026-08-29

  • Initial five-factor methodology.
  • Frozen absolute bands, goalposts, weights, coverage floors, required groups, aggregation rules, unavailable reasons, and rounding rules.
  • Added source-safe evidence provenance and an explicit SIPRI redistribution block.
  • Defined current preset and custom bloc validation contracts.
  • Froze per-input evidence ages and per-factor publication floors; stale or coverage-collapsed cohorts preserve the prior last-good snapshot.