"background": "The optimisation of manufacturing systems in developing economies requires robust, context-specific data on cost-effectiveness. However, methodological frameworks for generating such data through field trials in these settings are underdeveloped, limiting evidence-based investment and policy. ", "purpose and objectives": "This Data Descriptor presents the methodology and resultant dataset from a randomised field trial designed to evaluate the cost-effectiveness of interventions in manufacturing plant systems. The primary objective was to establish a replicable methodological framework for collecting high-fidelity engineering and economic data in an operational industrial context. ", "methodology": "A multi-site randomised controlled trial was implemented across functionally comparable manufacturing plants. The core statistical model for estimating the average treatment effect on the cost-efficiency ratio is specified as Y{it = \0 + \1 Tit + it^\ \ + \ +, where Yit is the log-transformed cost-efficiency metric for plant i at time t, Tit is the treatment assignment, and \ denotes plant-level fixed effects. Robust standard errors were clustered at the plant level. ", "findings": "The dataset comprises 12, 840 hourly observations across 24 key performance indicators, including energy consumption, throughput, and maintenance costs. Analysis indicates a statistically significant positive treatment effect on normalised output, with a point estimate suggesting an 8. 5% improvement (95% CI: 2. 1% to 14. 9%) under the implemented system interventions. ", "conclusion": "The trial successfully generated a novel, granular dataset that captures the interplay between technical system modifications and economic performance in a real-world manufacturing environment. The methodological approach proved feasible and generated data suitable for rigorous cost-benefit analysis. ", "recommendations": "Future research should apply this methodological framework to other industrial sectors and regions to build a comparative evidence base. Practitioners are encouraged to utilise the described dataset for benchmarking and validating simulation models
Jean de Dieu Niyonzima (Mon,) studied this question.