Multilevel Regression Analysis of Efficiency Gains in Ugandan Manufacturing Systems: A Policy Evaluation for Sustainable Industrialisation

{ "background": "Sustainable industrialisation in sub-Saharan Africa requires robust evidence on the efficiency of manufacturing systems to inform policy. Current evaluations often lack the methodological rigour to account for the hierarchical structure of plant-level data, where production units are nested within firms and sectors.", "purpose and objectives": "This policy analysis evaluates the application of multilevel regression modelling to measure efficiency gains in manufacturing plants. It aims to demonstrate the method's superiority in isolating system-level effects and to derive targeted policy insights for enhancing industrial productivity.", "methodology": "A three-level linear mixed-effects model was specified: $y{ijk} = \\beta0 + \\beta X{ijk} + u{k} + v{jk} + e{ijk}$, where $i$, $j$, and $k$ index production lines, plants, and sectors, respectively. The analysis used a novel, anonymised panel dataset of Ugandan manufacturing plants, with inference based on robust standard errors clustered at the sector level.", "findings": "The multilevel model revealed that 32% of the variance in technical efficiency was attributable to sector-level effects, a factor obscured by traditional ordinary least squares regression. A one-standard-deviation increase in managerial training investment at the plant level was associated with a 0.15 increase in efficiency score (95% CI: 0.11, 0.19).", "conclusion": "Multilevel regression provides a more accurate and policy-relevant framework for analysing manufacturing efficiency by quantifying the influence of different hierarchical levels. This approach is crucial for designing interventions that target the correct level of the industrial system.", "recommendations": "Industrial policy should shift from sector-agnostic support to targeted capacity building at the plant level, particularly in managerial training. National statistical offices should adopt hierarchical data structures to enable future multilevel analyses.", "key words": "multilevel modelling, industrial policy, manufacturing efficiency, sustainable development, sub-Saharan Africa", "contribution statement": "This article provides the first formal application of