Abstract
{ "background": "Water treatment systems in sub-Saharan Africa often operate below design capacity, leading to chronic water shortages. Systematic, quantitative evaluations of interventions to improve plant yield are scarce, hindering evidence-based asset management and investment.", "purpose and objectives": "This case study develops and applies a quasi-experimental analytical framework to rigorously quantify the causal impact of a major rehabilitation programme on the operational yield of selected water treatment works.", "methodology": "A difference-in-differences (DiD) model was employed, using panel data from treatment plants that underwent rehabilitation and a control group of similar, non-rehabilitated facilities. The core model is $Y{it} = \\beta0 + \\beta1 \\text{Treat}i + \\beta2 \\text{Post}t + \\delta (\\text{Treat}i \\cdot \\text{Post}t) + \\epsilon_{it}$, where $\\delta$ is the causal effect of interest. Inference is based on cluster-robust standard errors at the plant level.", "findings": "The rehabilitation programme significantly increased average daily yield. The DiD estimator $\\hat{\\delta}$ was 12.7 megalitres per day (95% CI: 8.4, 17.0), representing a 22% improvement relative to the pre-intervention mean for treated plants. The parallel trends assumption, tested using lead terms, was not violated.", "conclusion": "The applied DiD model provides a robust methodological framework for evaluating capital projects in civil engineering infrastructure, moving beyond simple before-after comparisons. The results confirm the efficacy of targeted rehabilitation in this context.", "recommendations": "Water authorities should adopt quasi-experimental evaluation techniques for post-project audits. Future rehabilitation programmes should prioritise the specific engineering interventions—particularly clarifier refurbishment and chemical dosing upgrades—identified as drivers of the yield gain.", "key words": "difference-in-differences, water treatment, infrastructure evaluation, causal inference, asset