MODEL UNCERTAINTY DISCLOSURE OBLIGATIONS

1. Meaning and Concept

Model uncertainty disclosure obligations in electricity regulation concern the duty of regulators, utilities, system operators and other decision-makers to identify and communicate material uncertainty contained in mathematical, economic, engineering or algorithmic models used for regulatory decisions. Modern electricity governance increasingly depends on models for demand forecasting, tariff calculation, renewable-output prediction, system adequacy, grid expansion, emissions assessment and investment planning.

A model does not perfectly reproduce reality. Its results depend upon assumptions, input data, scenarios and methodological choices. Consequently, regulatory decisions based on models should not present uncertain estimates as unquestionably accurate facts. In South African law, there is not a single general statute specifically titled a “model uncertainty disclosure” law. Rather, relevant obligations can arise through administrative-law rationality, procedural fairness, transparency, environmental assessment and sector-specific regulatory requirements.

2. Sources of Model Uncertainty

Electricity models may contain data uncertainty, where measurements or historical information are incomplete; parameter uncertainty, where variables such as future fuel prices or demand growth cannot be precisely predicted; and structural uncertainty, where the mathematical model simplifies the electricity system.

Scenario uncertainty is especially important in long-term electricity planning. Future demand, renewable-energy penetration, technology costs, carbon constraints and storage deployment may develop differently from assumptions embedded in a model.

Disclosure should therefore explain significant assumptions, limitations, data sources, sensitivity ranges and alternative scenarios where these factors materially affect the regulatory decision.

3. Administrative-Law Requirements

Under the Promotion of Administrative Justice Act 3 of 2000 (PAJA), administrative decisions may be reviewed where relevant considerations were ignored or decisions lack the required rational connection. This becomes important when a regulator relies heavily upon modelling.

A regulator does not necessarily have to eliminate uncertainty. Indeed, electricity regulation frequently requires decisions despite uncertainty. The legal concern is whether uncertainty has been reasonably recognised and incorporated into the decision-making process.

Disclosure also facilitates meaningful public participation. Stakeholders cannot effectively challenge tariff, generation or infrastructure assumptions if the fundamental methodology and limitations underlying the regulatory model remain inaccessible.

4. Case Law: Earthlife Africa Johannesburg v Minister of Environmental Affairs

Case Name/Citation: Earthlife Africa Johannesburg v Minister of Environmental Affairs and Others (65662/16) [2017] ZAGPPHC 58.

Facts: The case concerned environmental authorisation for the proposed Thabametsi coal-fired power station. Climate-change consequences, including greenhouse-gas emissions and the project's vulnerability to future climate conditions, had not been comprehensively assessed before authorisation.

Legal Issue: Whether the competent authorities had adequately considered climate-change impacts as relevant considerations before granting environmental authorisation.

Judgment: The High Court held that climate-change impacts were relevant considerations under the environmental regulatory framework. It required reconsideration after consideration of a climate-change impact assessment and public comments concerning that assessment.

Legal Principle/Ratio Decidendi: Administrative decision-makers must properly investigate and consider material information and relevant future impacts, even where legislation does not expressly prescribe one particular form of assessment. The Court observed that a formal expert climate report could provide the evidentiary basis for considering complex climate impacts.

Significance: Although not specifically a case about algorithmic model disclosure, Earthlife Africa supports the broader principle that uncertainty about future environmental and electricity-system conditions cannot simply be excluded from legally required decision-making.

5. Case Law: Sasol Oil v NERSA

Case Name/Citation: Sasol Oil (Pty) Ltd v National Energy Regulator of South Africa and Others (059715/2023) [2025] ZAGPPHC 919.

Facts: Sasol challenged NERSA's determination concerning petroleum pipeline tariffs and the application of the relevant pricing methodology.

Legal Issue: The dispute included whether NERSA had lawfully applied its regulatory methodology when determining tariffs.

Judgment: The High Court reiterated the principle derived from National Energy Regulator of South Africa v PG Group that a regulatory pricing methodology operates as a guideline rather than inflexible law, leaving regulatory discretion where rigid application could produce irrational or unlawful results.

Legal Principle/Ratio Decidendi: Regulatory methodologies must operate consistently with statutory requirements of rationality and, in the petroleum context considered there, principles including transparency, predictability and fairness.

Significance: Applied by analogy to electricity modelling, this demonstrates why regulators should not mechanically accept model outputs. They must evaluate assumptions and results against the governing statutory purpose and administrative-law standards.

6. Practical Disclosure Framework

A sound electricity-regulatory process should disclose the model's purpose, principal assumptions, input-data quality, uncertainty ranges, sensitivity analysis, alternative scenarios and material limitations. Where commercially confidential information cannot lawfully be published, sufficient methodological information should still be available to permit meaningful regulatory scrutiny.

7. Conclusion

Model uncertainty disclosure is increasingly important as electricity regulation becomes data-driven and algorithmically sophisticated. The governing principle is not that regulatory models must achieve perfect prediction. Rather, material uncertainty should be identified, evaluated and transparently incorporated into legally accountable decision-making. South African administrative and environmental case law provides a foundation for requiring regulators to consider relevant uncertainty rather than disguising uncertain projections as certain outcomes.