Skip to main content

Risk Intelligence

SARII Methodology

How the South Africa Risk Intelligence Index is constructed, interpreted, and limited

Overview

SARII v1.0 is a composite 0–100 risk score for South Africa. Higher scores indicate greater stress across four pillars: crime, fiscal, economic, and narrative.

The objective is transparency rather than black-box prediction. Each component is based on official or documented source data, normalised to a comparable 0–100 risk scale, then combined using fixed weights.

Pillar Weights

PillarWeightPrimary SourcePurpose
Crime30%SAPS crime statistics + Stats SA populationViolent crime risk using reported crime rates and population-normalised comparisons.
Fiscal25%National Treasury budget allocationsGovernment financial stress using debt, deficit, and expenditure pressure proxies.
Economic25%Stats SA employment, GDP, CPI inflationLabour market stress, economic output, and inflation pressure.
Narrative20%NewsAPI.org media coverage volumeInformation-environment volatility measured through unusual changes in article volume.

Composite Formula

SARII = (Crime × 0.30) + (Fiscal × 0.25) + (Economic × 0.25) + (Narrative × 0.20)

Each pillar is first normalised to a 0–100 risk score. The weighted sum produces the current composite reading displayed on the SARII dashboard.

Interpretation

0–25: low risk and broad stability.

25–50: moderate risk with pressure in selected pillars.

50–75: elevated risk with multiple stressed pillars.

75–100: high risk and broad deterioration.

Limitations

SARII is a directional indicator, not a prediction engine. It is intended to show the balance of stress across major systems, not to forecast specific political or economic events.

Fiscal measures rely partly on proxy variables in v1.0, and the narrative pillar should always be interpreted alongside data freshness and window length because media-volatility signals are sensitive to recent article volume.

The current methodology is calibrated for South Africa only. Scores should not be used for cross-country comparison without a dedicated normalisation framework.