Industrial machinery fleets in developing economies face significant operational risks, yet systematic, data-driven methodologies for quantifying and mitigating these risks are scarce. Existing approaches often rely on cross-sectional data, failing to capture temporal dynamics and unobserved heterogeneity. This study aims to develop and apply a panel-data econometric framework to evaluate risk factors and measure the effectiveness of targeted interventions for heavy machinery fleets in an industrial setting. A novel panel dataset was constructed from maintenance logs, incident reports, and operational records for a fleet of 87 units across multiple sites. A fixed-effects model was estimated: Risk₈ₓ = ᵢ + ₁ Intervention₈ₓ + ₂ Utilisation₈ₓ + X₈ₓ + ₈ₓ, where ᵢ captures unit-specific heterogeneity. Inference is based on robust standard errors clustered at the site level. The implemented preventive maintenance intervention was associated with a 22. 5% reduction in major incident risk (95% CI: 18. 1% to 26. 9%). Furthermore, a non-linear relationship was identified, where risk increased disproportionately with utilisation rates beyond a specific threshold. The panel-data approach provides a robust methodological advancement for fleet risk analysis, confirming that structured maintenance protocols are highly effective. The model successfully isolates the intervention effect from time-invariant unobserved confounders. Fleet managers should adopt panel-data methodologies for continuous risk monitoring. Policies should mandate the integration of such analytical frameworks into national equipment management guidelines to institutionalise evidence-based practice. asset management, panel data, fixed effects, risk modelling, maintenance engineering, industrial safety This paper provides the first application of a panel-data econometric model to quantify risk reduction in industrial machinery fleets within a West African context, establishing a replicable methodology for the region.
Kwame Gyeabour Asante (Wed,) studied this question.