Abstract
{ "background": "The assessment of cost-effectiveness for industrial machinery fleets in developing economies is often hampered by sparse, heterogeneous data and the need to integrate operational parameters with economic constraints. Existing deterministic models frequently fail to quantify the uncertainty inherent in such complex engineering systems.", "purpose and objectives": "This working paper presents a methodological evaluation of a novel Bayesian hierarchical model designed to rigorously measure the cost-effectiveness of industrial machinery fleets. The objective is to provide a robust framework that quantifies uncertainty and pools information across disparate fleet units.", "methodology": "We develop a three-level hierarchical model where the cost-effectiveness ratio for the $i^{th}$ machine is modelled as $\\text{CER}i \\sim \\text{Normal}(\\thetai, \\sigma^2)$, with $\\theta_i \\sim \\text{Normal}(\\mu, \\tau^2)$ representing fleet-level heterogeneity. Hyperpriors are placed on $\\mu$ and precision parameters. Inference is performed using Hamiltonian Monte Carlo, with model performance evaluated via posterior predictive checks and comparison to a non-hierarchical baseline.", "findings": "The methodological evaluation demonstrates that the hierarchical structure effectively shrinks estimates for under-reported units towards the group mean, reducing extreme and implausible point estimates. A key finding is that the 95% credible interval for the fleet-wide mean cost-effectiveness ratio was approximately 40% narrower under the hierarchical model compared to the non-pooled alternative, indicating substantially increased precision.", "conclusion": "The proposed Bayesian hierarchical model offers a statistically rigorous methodology for cost-effectiveness analysis in contexts with fragmented data. It formally incorporates uncertainty and improves estimation efficiency through partial pooling, providing a superior alternative to conventional deterministic or fully disaggregated approaches.", "recommendations": "Adoption of this modelling framework is recommended for asset management and procurement planning within the industrial sector. Future work should focus on integrating real-time sensor data into the model's observational layer and developing user-friendly software implementations for practitioners.", "key words": "Bayesian inference,