Skip to main content
The five-factor scorecard answers five structural questions for a country or bloc: can it feed itself, supply its energy, sustain its people, develop and use technology, and defend itself? It is an annual-cadence judgment layer over source observations. It is not the Country Instability Index (CII), the Country Resilience Index (CRI), or a replacement for either one.

​
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. A change to a weight, goalpost, band, coverage floor, required group, aggregation rule, or input mapping requires a methodology version and changelog entry. A storage shape change also requires a schema version change.

​
Common scoring rules

Each input is normalized to an absolute 0-100 component score with the frozen goalposts below. Values outside a goalpost are clamped. The scorer then:
  1. Adds the declared weights of available components.
  2. Reports that sum as inputCoverage from 0 to 1.
  3. Divides the available weighted component total by available weight.
  4. Applies the pillar coverage floor and required-group rules.
  5. Maps the unrounded continuous sub-score to an integer band.
  6. Rounds the published sub-score and coverage to two and four decimal places, respectively. Rounding never changes the band, because the band is always derived in step 5 from the unrounded sub-score. Near a boundary the two can therefore disagree by design: a continuous sub-score of 19.996 publishes as subScore: 20.00 with band: 1, not band: 2. Read the published band; do not re-derive it from the published subScore.
Unavailable inputs do not receive a neutral score and do not enter the weighted mean. Internally, a pillar below its floor has hasScore=false, score=null, subScore=null, and one or more explicit reasons. Protobuf JSON keeps numeric fields present as zero; API and MCP clients must read hasScore first because those zeros are insufficient-data placeholders, not measured zero resilience.

​
Absolute bands

The lower boundary is inclusive. For example, exactly 40.0 is band 3.

​
Unavailable reasons

The closed reason set is:
  • source-unavailable: the upstream snapshot could not be read.
  • country-unavailable: the source is healthy but has no country observation.
  • invalid-value: the observation is non-finite or outside its valid domain.
  • stale: the observation is older than the input freshness contract.
  • coverage-below-floor: available weight is below the pillar floor.
  • required-group-missing: no available input satisfies a required group.
  • missing-population: a population-weighted calculation has no population.
  • redistribution-blocked: source policy forbids the partner-facing evidence.
Each evidence item is a tagged available/unavailable record. Available evidence includes input ID, value, year, unit, source, and source key. A retained last-good upstream observation stays available and carries quality=retained.

​
Freshness and publication floors

V1 applies observation-age limits before scoring. Population, food-balance, food-stock, and age-structure inputs allow at most 3 years; physical energy balance allows 4; low-carbon generation, trade diversity, workforce, and defense allow 5; all other annual inputs allow 7. The boundary year is inclusive. An expired source content-age envelope also makes its declared inputs stale even when the stored values remain readable. A daily cohort replaces last-good only when at least 180 countries have one scoreable pillar, at least 150 have usable population evidence, and the scoreable-country counts meet these frozen floors: food 80, energy 120, demographics 150, technology 120, and defense 30. The floors were set from the 196-country production-source audit with headroom for normal source coverage. Health publishes population and the five pillar counts separately. A partial source outage cannot refresh health or replace a richer cohort.

​
Food

Food measures the ability to meet caloric use from production and stocks. It uses commodity production, consumption, ending stocks, and use from resilience:food-stocks:v1. Source quantities in thousand metric tonnes are converted to trillion kcal as thousand metric tonnes * kcal/kg / 1,000,000 with the frozen commodity conversion table before aggregation. Coverage floor: 0.70. Calorie production / consumption is required.
  • Band 1: production is near or below half of use and buffers are weak.
  • Band 3: domestic output covers much, but not all, use or buffers are mixed.
  • Band 5: output materially exceeds use and stock buffers are strong.
For blocs, production and consumption are summed before their ratio is scored; ending stocks and total use are also summed before scoring. Water and import diversity component scores are population-weighted across members with evidence.

​
Energy

Energy measures physical primary-energy balance and supporting power-system quality. Consumption comes from OWID primary_energy_consumption. Production is derived from World Bank net energy imports (EG.IMP.CONS.ZS):
Coverage floor: 0.60. Primary production / consumption is required.
  • Band 1: production covers little of use.
  • Band 3: balance is mixed and imports remain material.
  • Band 5: the country is energy self-sufficient or a net producer with strong supporting power-system evidence.
For blocs, primary production and consumption TWh are summed before scoring. Low-carbon generation and grid efficiency component scores are population-weighted across members with evidence.

​
Demographics

Demographics combines age structure, the projected working-age population, education, research capacity, and trained industrial labor. 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.
  • Band 1: severe dependency or contraction with little capability evidence.
  • Band 3: mixed age structure and human-capability depth.
  • Band 5: favorable labor supply and deep education, research, and industrial workforce capacity.
Bloc demographics is the population-weighted mean of member unrounded continuous demographic sub-scores. It never averages integer bands.

​
Technology

Technology uses source-preserving World Bank observations. The existing technology readiness score, rank, and components remain 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.
  • Band 1: limited digital access and little measured innovation capacity.
  • Band 3: broad use or research capacity, but material gaps remain.
  • Band 5: high connectivity plus deep and sustained innovation capability.
Bloc technology is the population-weighted mean of member unrounded continuous technology sub-scores.

​
Defense

Defense measures sustained national defense capacity. World Bank defense observations come from military:industrial-base:v1. SIPRI raw transfer rows are not stored or returned. Until the source policy permits public use of the already derived supplier HHI, that component is explicitly unavailable with redistribution-blocked. Coverage floor: 0.50. At least one posture input (spending or personnel) and the arms-transfer industrial-balance input are required.
  • Band 1: small posture with high external equipment dependence.
  • Band 3: material posture or industrial capability, with important gaps.
  • Band 5: large sustained posture and strong domestic/export industrial depth.
Bloc defense is the population-weighted mean of member unrounded continuous defense sub-scores. Supplier diversity remains unavailable while the policy block applies. The industrial balance is available only when both exports and imports are explicit finite observations from the same year. A missing side is never coerced to zero; an explicit measured zero remains valid.

​
Bloc rules

V1 presets are versioned as part of the methodology: USMCA, EU27, BRICS, GCC, ASEAN, and NATO. Custom blocs accept 2-30 unique uppercase ISO-2 country codes from the public rankable universe. A request must select exactly one preset or one custom member list. Presets may contain more than 30 members. Membership was checked 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. Preset membership changes require a methodology changelog entry. Each bloc pillar reports its aggregation method, included and excluded members, population coverage where applicable, and the evidence used. Food and energy use physical aggregate-then-score rules. Demographics, technology, and defense population-weight continuous scores. No bloc formula averages country bands.

​
Changelog

​
1.0.0 - 2026-08-29

  • Initial five-pillar 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-pillar publication floors; stale or coverage-collapsed cohorts preserve the prior last-good snapshot.