"background": "Previous engineering-economic analyses of centralised water treatment infrastructure in sub-Saharan Africa have produced inconsistent conclusions regarding long-term cost-effectiveness. These discrepancies hinder evidence-based policy and investment, suggesting potential methodological limitations in prior modelling approaches. ", "purpose and objectives": "This replication study conducts a methodological diagnostic of prior cost-effectiveness models. Its primary objective is to verify and extend the original multilevel regression analysis using an expanded dataset and robust statistical diagnostics to assess the reliability of earlier findings. ", "methodology": "We executed a direct replication followed by a measurement-error-in-variables analysis. The core multilevel model is specified as Cost{ij = \0j + \1Volumeij + \2Ageij + u0j + eij, where j denotes the facility. We employed cluster-robust standard errors and conducted a full suite of sensitivity analyses, including tests for heteroscedasticity and influential outliers. ", "findings": "The replication revealed a significant upward bias in earlier cost-effectiveness estimates. Specifically, the original coefficient for treatment volume was overestimated by approximately 22%. Our diagnostic identified that failure to account for non-linear depreciation of mechanical components was a key source of this bias, materially affecting the projected net present value of systems. ", "conclusion": "The original study's conclusions on cost-effectiveness are not robust to methodological refinement. The significant bias underscores the critical importance of incorporating engineering degradation parameters into economic models of infrastructure. ", "recommendations": "Future engineering-economic analyses of public infrastructure should integrate physical asset degradation models into their regression frameworks. Policymakers should treat cost projections from models lacking such diagnostics with caution. ", "key words": "replication study, infrastructure economics, multilevel modelling, water treatment, cost-benefit analysis, asset management", "contribution statement": "This study provides a novel diagnostic framework that integrates engineering degradation rates into regression-based cost-effectiveness models, yielding a more robust tool for infrastructure investment appraisal
Mwinyi et al. (2009) studied this question.