Artificial Intelligence Governance of National Health Insurance Benefit Package Algorithms Under Unmeasured Parliamentary Budget Committee Reopening Authority: A Political Fairness Estimation

The problem is not algorithmic error but institutional: the reopening power is a political variable that is neither observed nor modelled in existing AI governance frameworks, and its exercise can systematically redistribute fairness away from the actuarial and clinical criteria encoded in benefit-package algorithms. Using a stylised model of committee preferences, algorithmic rule-making and stakeholder contestation, the analysis identifies three mechanisms—reopening frequency, reopening scope and reopening timing—through which unmeasured parliamentary authority degrades procedural and distributive fairness. The framework is then applied illustratively to the institutional design of AI governance for national health insurance in lower- and middle-income African contexts, where parliamentary budget committees often combine appropriation and oversight functions. Fairness estimation for benefit-package algorithms must incorporate institutional parameters that are currently treated as exogenous noise, and it proposes a governance architecture that makes reopening authority observable, bounded and contestable. The contribution is a political-science reframing of AI fairness as an institutional estimation problem rather than a purely technical one.