Abstract
{ "background": "Operational risk in manufacturing systems is a critical engineering management concern, particularly in developing industrial contexts. Previous studies have proposed panel-data models for risk quantification, yet their methodological robustness and replicability in specific regional settings require rigorous evaluation.", "purpose and objectives": "This replication study aims to methodologically evaluate a published panel-data approach for measuring operational risk reduction in manufacturing plant systems. The core objective is to assess the model's specification, estimation stability, and practical applicability within an industrialising context.", "methodology": "We replicate the econometric analysis using an extended, proprietary dataset from multiple manufacturing facilities. The core model is a two-way fixed effects specification: $Risk{it} = \\alpha + \\beta1 System{it} + \\beta2 X{it} + \\mui + \\lambdat + \\epsilon{it}$, where $Risk_{it}$ is a composite operational risk index. Estimation employs Driscoll-Kraay standard errors to account for cross-sectional dependence and heteroskedasticity.", "findings": "The replication confirms the original study's central finding that integrated plant systems are associated with reduced operational risk, but the effect magnitude is smaller. Specifically, the coefficient for the primary system variable was approximately 40% lower than originally reported. Statistical inference showed the relationship remained significant at the 5% level, with a 95% confidence interval that did not include zero.", "conclusion": "The panel-data approach provides a valid framework for risk assessment, but effect sizes are sensitive to dataset composition and error structure assumptions. This underscores the importance of robust estimation techniques in engineering management applications.", "recommendations": "Future applications of this model should mandate diagnostic testing for spatial and serial correlation. Practitioners should calibrate risk reduction expectations based on localised, context-specific re-estimations rather than transferring previously published coefficients directly.", "key words": "replication study, operational risk, panel data, manufacturing systems, econometric evaluation, industrial engineering", "contribution