> ## Documentation Index
> Fetch the complete documentation index at: https://worldmonitor-spike-bun-package-manager.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Resilience Indicators

> The indicator-level explainability layer under the country resilience score — normalized values, runtime weights, contributions reconciled per dimension, observation age, and source provenance.

## Start here

**This is the receipt.**

[The country resilience index](/methodology/country-resilience-index) gives a
country one number. This gives you every one of the 72 ingredients behind it:
what each indicator scored, how much it counted, and which organization
published it.

Use it when you need to answer "*why* did this country move?" — because the
answer is always a specific indicator, and this is where it is named. It is also
what makes the index arguable: you can disagree with a particular indicator's
contribution only if you can see it.

### Two things to check before trusting a row

**Read `state` before any number.** An indicator can be `observed`, `imputed`,
`missing`, `fallback`, `not-applicable`, `source-failure`, `inactive`, or
`retired`. An imputed or fallback value can still be a legitimate part of the
score — but it is an inference, not a measurement, and the field says which.

**A blank raw value is usually a licence, not a gap.** Where an upstream
publisher forbids redistribution, the row serves WorldMonitor's normalized score
and the attribution but withholds the underlying number. The engine has the
value and used it; it just may not republish it. See
[Raw values and licensing](#raw-values-and-licensing).

### Why weights look different from the published defaults

The weights served here are the ones **actually applied on this run**, not the
registry defaults. When an indicator is unavailable for a country, the engine
redistributes its weight across the rest. Reading the default instead would
misattribute the score.

<Info>
  **Access.** Served from `GET /api/resilience/v1/get-resilience-indicators?countryCode=<ISO2>` and MCP `get_resilience_indicators`. Both require a Pro subscription. There is no dashboard surface — this is an API and agent endpoint.
</Info>

## Per indicator

Each of the 72 rows carries:

| Field              | Meaning                                                                                                                                  |
| ------------------ | ---------------------------------------------------------------------------------------------------------------------------------------- |
| Normalized score   | The indicator on the engine's common scale                                                                                               |
| State              | `observed`, `imputed`, `missing`, `fallback`, `not-applicable`, `source-failure`, `inactive`, or `retired`                               |
| Runtime weight     | The weight actually applied on this run, not the registry default                                                                        |
| Contribution       | The indicator's share of its dimension, reconciled so the dimension's contributions sum correctly                                        |
| Availability flags | `normalizedScoreAvailable`, `runtimeWeightAvailable`, and `scoringWeightShareAvailable` decide whether the matching numeric fields apply |
| Observation age    | How old the underlying observation is                                                                                                    |
| Source provenance  | Which upstream published it                                                                                                              |

Weights are served as applied rather than as configured because the engine renormalizes when an indicator is unavailable for a country. Reading the registry default would misattribute the score.

## Indicator states

`observed` and `imputed` are only two possible states. `missing` means no usable observation resolved, `fallback` means a lower-confidence substitute was used, `not-applicable` is structural, `source-failure` is an upstream or seed failure, and `inactive` or `retired` marks an indicator outside the active scoring path. An imputed or fallback indicator can still be a real part of the score, but it is not a measurement. Read `state` and the availability flags before any numeric field.

## Raw values and licensing

Raw source values are included **only where the upstream license permits redistribution**. Elsewhere the normalized score and the provenance are served without the underlying number.

That absence is a licensing boundary, not missing data — the engine has the value and used it; WorldMonitor may not republish it. The per-source terms are recorded in [Resilience indicator licensing](/methodology/resilience-indicator-licensing).

## Reconciliation

Contributions are reconciled to each dimension rather than reported as raw products. A caller can sum the contributions within a dimension and get the dimension's score; summing across dimensions with their dimension weights gives the country score. Values that did not reconcile are not silently rounded into place — the indicator reports its state instead.

## Related

* [Pro Intelligence Suite](/pro-intelligence-suite)
* [Country resilience index](/methodology/country-resilience-index) — the score this explains
* [Resilience indicator licensing](/methodology/resilience-indicator-licensing)