"background": "Municipal infrastructure systems in many developing nations face chronic reliability challenges, yet policy evaluation often lacks robust, quantitative frameworks that account for hierarchical data structures. Existing assessments rarely disentangle asset-level performance from systemic, municipality-wide factors. ", "purpose and objectives": "This policy analysis evaluates the application of multilevel regression modelling to measure and predict the reliability of municipal infrastructure systems. It aims to demonstrate how this methodology can isolate policy-relevant variance at different administrative levels to inform targeted interventions. ", "methodology": "A multilevel logistic regression model is constructed, nesting individual infrastructure assets within municipalities. The core model is specified as \ (p{ij) = \0j + \ Xij with \0j = \00 + u0j, where pij is the reliability probability for asset i in municipality j, Xij are asset-level covariates, and u₀₉ are municipality-level random effects. Analysis uses a synthesised national dataset of water, road, and drainage assets. Inference is based on robust standard errors. ", "findings": "Municipality-level random effects account for approximately 30% of the unexplained variance in asset reliability, indicating that systemic governance and fiscal factors are substantial drivers. A key concrete result is that a one-unit increase in a municipality's technical capacity index is associated with a 15% higher odds of asset reliability (95% CI: 5% to 26%), holding asset age and type constant. ", "conclusion": "Multilevel regression provides a statistically rigorous framework for policy analysis, effectively partitioning variance to reveal where interventions—asset-specific upgrades versus municipal capacity building—would be most impactful. It moves beyond descriptive failure rates to model causal pathways. ", "recommendations": "National infrastructure policy should mandate the collection of hierarchically structured asset data. Investment decisions should be informed by variance decomposition analyses to allocate resources between direct asset rehabilitation and strengthening municipal institutional capacities. ",
Nalwanga et al. (Thu,) studied this question.